Sign in with
Sign up | Sign in
Your question
Closed

Radeon HD 5970 Scaling

Tags:
Last response: in CPUs
Share
a b à CPUs
December 18, 2009 4:21:22 AM

I think many of you remember this article here: http://www.legionhardware.com/document.php?id=869

The article was brought up up during a bitter and vigorous debate here on THG but I figure it deserves mention. Over the course of that debate I had noticed a trend, a trend which was not entirely obvious when looking at the results on their own. So what one of my friends Tactician did was take it upon himself to piece together the data in order to bring to light a clear picture.

His Original post can be seen here: http://tinyurl.com/ycc8wd4

Enjoy :D 

Intel = RED
AMD = ORANGE


What I see here is that with each OC bump there is an increase in FPS for both CPUs. More MHz = more FPS ... okay I get that.


Here I see very minimal FPS increase with more CPU power. Also as far as I can tell these CPUs are on Par.


Lower clocks i7 has more FPS, higher clocks pretty much the same. After 2.6GHz there isn't much gain in FPS for the extra MHz


Here there is quite a noticeable difference up until 3.6GHz. The i7 hit a wall at 2.6GHz and there was no benefit from additional OCing.


Again at the low end of the GHz scale a distinct difference. Both CPUs cease to receive any benefit from MOAR POWAH! at approximately the 3.0GHz mark.


First thing I notice... DAMN really low FPS across all tests. Second thing: another brick wall.


In my opinion this shows no improvement from OCing at all. AMD is a stride ahead at the 2.8GHz mark.


AMD receives a small boost from OCing and steps in front of intel at 3.6GHz


Pretty much starts off infront of the wall here. AMD picks up a few points at the end of the race.

That is a lot of information to process but what you're seeing is that the Core i7 Processor IS the faster gaming CPU (look at it's performance at 2GHz compared to the Phenom II X4). As you increase the clock speed, however, a Graphics bottleneck occurs. The perfect example is the last shot. The GPU becomes the bottleneck even at 2GHz. The Core i7 stops scaling from there.

What many users are quick to point out (particularly those who favor AMD) is that AMD sometimes manager to get as much as ~3FPS more than Intel when a GPU bottleneck is present. These users mistakenly assume this is some sort of proof that "AMD is the better gaming chip offering better performance" as one user suggested. This assertion is patently false. By objectively looking at the data we can come to a rather simple Hypothesis. If one were to alleviate the Graphics Card bottleneck by, let's say, adding another Graphics card... the Core i7 would out-scale the Phenom II X4 (we can deduce this by looking at how many more clocks it takes a Phenom II X4 before it can match a Core i7 running at only 2GHz).

This leaves us with the following question... "Why is the Phenom II X4 out performing Intel by a few frames during instances when a GPU bottleneck occurs?".

One could hypothesize that since this ONLY occurs in instances where we're using a GPU that it has something to do with CPU to GPU communications. There is a bottleneck or a performance inhibitor at that precise area. What could it be?

Well I think I've found the answer and have tried to painstakingly spread the answer in order to further knowledge. Intel holds a patent which suggest (clearly states actually) that there is an inherent incompatibility between the communications protocol the PCI Express bus uses and that of the QPi link. This patent explains how the IOH (X58 chip) acts as a hub converter and "tunnels" the data through (in one cycle). This "cycle" means latency in lay mans terms. Where is this patent you ask? Here: http://www.patentstorm.us/patents/7210000/description.h...

And there is your answer. :) 

I would like to thank Tactician for piecing these graphs together as well as the crew over at Legion Hardware for the raw data.

Peace.

More about : radeon 5970 scaling

December 18, 2009 5:06:40 AM

I thought the scaling looked a bit more optimistic for the i7 9xx compared to the Phenom II X4 and realised this is from HD5970 which is dual-GPU (a niche market).
Of course there's no denying which is the superior and longer-lasting gaming platform in the long run regarding CPU-GPU scaling balance, but thought people should take notice when reading those data before deciding which is the right platform for their need.

The data shows the next gen. top single-GPU (usually 65-80% increase from current, 1-1.5yrs from now) the Phenom II X4 @3.6Ghz+ won't become a bottleneck @2560x1600.
a b à CPUs
December 18, 2009 5:09:47 AM

well structred post, but anyone who says AMD is clearly better si simply being a fanboy.

intel processors are better, pure and simple. but AMD are better value.

to me, that is why AMD procs are better for any build on a remotely limited budget. lose 2 or 3 frames per second by geting an am3 proc, but save enough money to get a much better vdeo card. and gain 10=20 FPS.

if money were no option, id be building intel. but by going amd instead, i got to have a 4890 instead of a 4850 in my rig. that sure as hell means the AMD rig gets better performance.
Related resources
a b à CPUs
December 18, 2009 5:19:18 AM

When it comes to pure gaming (and nothing else) the gamers CPU of choice would be one that offers you value and performance.

In that case it's a tough choice because there are two CPUs available which offer a good Price/Performance ratio.

I am of course talking about the Phenom II X4 955/965 and Corei5 750. Either of these two would give a gamer top notch performance.

It's when we move outside of the realm of gaming that the Phenom II X4 begins to lose it's luster. In terms of pure processing performance the Corei5/i7 Processors offer more kick.

It's up to the buyer to determine what is within their budget and structure a purchase based on a balance between need and funds available.

As of the time of writing the best deal going (assuming you live near a Microcenter) would be the Corei5 750 for $149USD. The rest of the system (motherboard and RAM) is identical to the equivalent AMD based alternatives.

When the Corei5 is sold closer to it's MRSP the tables turn and the Phenom II X4 955 becomes the better alternative in terms of Price/Performance for the gamer.
a c 123 à CPUs
December 18, 2009 6:03:25 AM

wuzy said:
I thought the scaling looked a bit more optimistic for the i7 9xx compared to the Phenom II X4 and realised this is from HD5970 which is dual-GPU (a niche market).
Of course there's no denying which is the superior and longer-lasting gaming platform in the long run regarding CPU-GPU scaling balance, but thought people should take notice when reading those data before deciding which is the right platform for their need.

The data shows the next gen. top single-GPU (usually 65-80% increase from current, 1-1.5yrs from now) the Phenom II X4 @3.6Ghz+ won't become a bottleneck @2560x1600.


The 5970 may be a dual GPU card but the way it works is completely different. The Crossfire does not work the same as if you have two 5850s. It also has to share one PCIe 2.0 x16 slot instead of each having their own PCIe 2.0 x16 slot. To me this may mean that there is not enough bandwidth for both GPUs to pull out of the graphics bottleneck.

But still if you do add in another 5970 I would be pretty sure that a Core i7 would start to pull away and it would take more GHz from a Phenom II X4 to match a 2GHz Core i7.

Then again since this is contrary to all belief that AMD iz da godz, it is wrong........
a b à CPUs
December 18, 2009 7:51:43 AM

Its great that they used all the various clock speeds for both cpus for this test. It shows exactly what it is intended to show. Cpu scaling.


But it in reality nobody should care what the i7 or the x4 does below stock speeds. So the performance at 2Ghz really means nothing.
a b à CPUs
December 18, 2009 9:47:43 AM

ElMoIsEviL said:
When it comes to pure gaming (and nothing else) the gamers CPU of choice would be one that offers you value and performance.

In that case it's a tough choice because there are two CPUs available which offer a good Price/Performance ratio.

I am of course talking about the Phenom II X4 955/965 and Corei5 750. Either of these two would give a gamer top notch performance.

It's when we move outside of the realm of gaming that the Phenom II X4 begins to lose it's luster. In terms of pure processing performance the Corei5/i7 Processors offer more kick.

It's up to the buyer to determine what is within their budget and structure a purchase based on a balance between need and funds available.

As of the time of writing the best deal going (assuming you live near a Microcenter) would be the Corei5 750 for $149USD. The rest of the system (motherboard and RAM) is identical to the equivalent AMD based alternatives.

When the Corei5 is sold closer to it's MRSP the tables turn and the Phenom II X4 955 becomes the better alternative in terms of Price/Performance for the gamer.


I think you are mostly right but I would like to add something. The Phenom II X4 and Core i5 750 processors perform extremely close with the Core i5 750 pulling slightly ahead in some areas and it has some nice features like both SLI and CFX support on one board, but it costs more. Whether that is worth it to you or not is totally dependant on the person. I went with a Core i5 750 and bought mine at Tigerdirect for $190 shipped. I did this because the ability to SLI and CFX on the same board is very important to me because I absolutely hate nVida chipsets.
December 18, 2009 10:42:51 AM

jimmysmitty said:
The 5970 may be a dual GPU card but the way it works is completely different. The Crossfire does not work the same as if you have two 5850s. It also has to share one PCIe 2.0 x16 slot instead of each having their own PCIe 2.0 x16 slot. To me this may mean that there is not enough bandwidth for both GPUs to pull out of the graphics bottleneck.

But still if you do add in another 5970 I would be pretty sure that a Core i7 would start to pull away and it would take more GHz from a Phenom II X4 to match a 2GHz Core i7.

Then again since this is contrary to all belief that AMD iz da godz, it is wrong........


The point was SLi/CrossFire being more CPU intensive, on top of the usual CPU-GPU scaling (should be common knowledge by now). Nothing to do with PCIe bandwidth.
Examples of that are HD4870X2 vs. HD5870 or 9800GX2 vs. GTX280 (and soon GTX295 vs. Fermi).

Afterall, SLi and Crossfire is all done purely in software with zero relevance to hardware, other than firmware license bitching by nVidia.
Anonymous
a b à CPUs
December 18, 2009 11:13:18 PM

Is that reliable?

http://www.xbitlabs.com/articles/video/display/radeon-h...

Quote:
It was harder with the AMD Phenom II X4 965 processor because my mainboard could not increase the base frequency higher than 256MHz.


Quote:
Despite my using fast DDR2 modules in the system, the integrated memory controller of the AMD processor imposed some limitations resulting in a performance reduction.


Quote:
The use of different system memory, DDR3 for the Intel Core i7 platform and DDR2 for the AMD Phenom II X4 platform, makes it impossible to compare these two platforms directly.


Quote:
One more pitfall for the AMD platform in this test session was PCI Express. The Gigabyte GA-MA790GP-DS4H mainboard is based on the AMD 790GX chipset which supports two ATI Radeon cards in CrossFireX only as PCIe x16 + PCIe x8, as opposed to both platforms from Intel which support two PCIe x16.


It would seem to be stacked against the amd system from the start. the xbitlabs article cant be compared to the other one which was made with both systems at their maximums. in this one, the amd is more like a midrange system instead of toprange.
a b à CPUs
December 18, 2009 11:42:40 PM

Quote:
Is that reliable?

http://www.xbitlabs.com/articles/video/display/radeon-h...

Quote:
It was harder with the AMD Phenom II X4 965 processor because my mainboard could not increase the base frequency higher than 256MHz.


Quote:
Despite my using fast DDR2 modules in the system, the integrated memory controller of the AMD processor imposed some limitations resulting in a performance reduction.


Quote:
The use of different system memory, DDR3 for the Intel Core i7 platform and DDR2 for the AMD Phenom II X4 platform, makes it impossible to compare these two platforms directly.


Quote:
One more pitfall for the AMD platform in this test session was PCI Express. The Gigabyte GA-MA790GP-DS4H mainboard is based on the AMD 790GX chipset which supports two ATI Radeon cards in CrossFireX only as PCIe x16 + PCIe x8, as opposed to both platforms from Intel which support two PCIe x16.


It would seem to be stacked against the amd system from the start. the xbitlabs article cant be compared to the other one which was made with both systems at their maximums. in this one, the amd is more like a midrange system instead of toprange.

If you can find me a test which shows DDR3 memory boosting Phenom II X4 performance under games.. I will give you that point.

As for the limitations on the AMD platform under overclocked scenarios. These are the case for ALL AMD platforms. Trying to get 1T working under overclocked conditions is near impossible on an AMD platform (without incurring the random blue screen). I have attempted to do so on a 790x platform with little luck and many people have issues running their memory over 1333MHz on overclocked AMD platforms (not everyone but many people do).

As for the PCIe mention. They're all 2.0 variants and PCI Express x8 2.0 is by no means a bottleneck for a single Radeon HD 5870 (it is equal in bandwidth to a PCIe x16 v 1.1 slot).

But even if all of the above was removed from the equation.. there is no way the AMd Platform could make up for the performance deficit. It performs on par with a Core i7 @ 2.7GHz (Phenom II X4 @ 4.1GHz that is) several times under conditions when there is no GPU bottleneck.

This is not surprising as it is the case in CPU only applications such as Video encoding, 3D Rendering, Folding@Home and several other tasks.
Anonymous
a b à CPUs
December 19, 2009 12:24:41 AM

but it's not just about the memory or the pcie bandwith or the overclocking issues, its about them all hurting this amd system at the same time.

if we could find a p55 review that blew up the motherboard would that make the core i5 worse than the phenom 2? that wouldnt be considered to be a fair test and neither should this one and they said as much on xbitlabs too.
a b à CPUs
December 19, 2009 12:39:36 AM

Quote:
but it's not just about the memory or the pcie bandwith or the overclocking issues, its about them all hurting this amd system at the same time.

if we could find a p55 review that blew up the motherboard would that make the core i5 worse than the phenom 2? that wouldnt be considered to be a fair test and neither should this one and they said as much on xbitlabs too.

Fair or not the results wouldn't be any different.

This is how both these platforms operate.

Using DDR2 or DDR3 actually makes no difference in performance for the Phenom II X4. So that argument is negligible.

The fact that AMDs memory controller has issues at 1T timing when highly overclocked is a part of the AMD platform so it's fair. The only other aspect you mentioned was the PCIe x8 2.0 slot used (x16 + x8) which really makes no difference in terms of performance:


Here give the AMD platform 1FPS more at those settings (because it's using x16 and x8). It won't change the results seen above (and the 3FPS boost was garnered with a Core i7 920 @ 3.8GHz).

It really make no difference because none of these change the totality of the results. You're not going to get the 60FPS more required to match a Core i7 Under Far Cry 2 for example (by making these changes).
a c 123 à CPUs
December 19, 2009 5:16:05 AM

wuzy said:
The point was SLi/CrossFire being more CPU intensive, on top of the usual CPU-GPU scaling (should be common knowledge by now). Nothing to do with PCIe bandwidth.
Examples of that are HD4870X2 vs. HD5870 or 9800GX2 vs. GTX280 (and soon GTX295 vs. Fermi).

Afterall, SLi and Crossfire is all done purely in software with zero relevance to hardware, other than firmware license bitching by nVidia.


Well its all software when its two GPUs in CG/SLI. With the x2 GPUs it turns more hardware based.

My only meaning is that its possible for both GPUs to be unhappy sharing one x16 link but probably not for a few more gens.

Quote:
Is that reliable?

http://www.xbitlabs.com/articles/video/display/radeon-h...

Quote:
It was harder with the AMD Phenom II X4 965 processor because my mainboard could not increase the base frequency higher than 256MHz.


Quote:
Despite my using fast DDR2 modules in the system, the integrated memory controller of the AMD processor imposed some limitations resulting in a performance reduction.


Quote:
The use of different system memory, DDR3 for the Intel Core i7 platform and DDR2 for the AMD Phenom II X4 platform, makes it impossible to compare these two platforms directly.


Quote:
One more pitfall for the AMD platform in this test session was PCI Express. The Gigabyte GA-MA790GP-DS4H mainboard is based on the AMD 790GX chipset which supports two ATI Radeon cards in CrossFireX only as PCIe x16 + PCIe x8, as opposed to both platforms from Intel which support two PCIe x16.


It would seem to be stacked against the amd system from the start. the xbitlabs article cant be compared to the other one which was made with both systems at their maximums. in this one, the amd is more like a midrange system instead of toprange.


The difference between DDR2 and DDR3 is not enough to warrant anything. When it comes to that the only advantage DDR3 will have is in another year or so when it is able to get higher frequencies and lower timings. Currently DDR3 1600 in most cases produces the same performance as DDR2 800.

And as said, PCIe x8 is fine for most GPUs. It probably will be for a few more gen of GPUs. Of course they don't want you to know that so you think you need the latest GPU interface but in reality even with PCIe 1.1 x6 vs PCIe 2.0 x16 there is such a marginal loss (I mean in like the 1 FPS or lower) , even if one system was using PCIe 1.1 x16 it wouldn't be that unfair of an advantage.
Anonymous
a b à CPUs
December 19, 2009 7:46:38 AM

the amd motherboard is 8x 8x not 16x 8x there is no such thing as 16x 8x, its either 16x 1x or 8x 8x. Two 5870 at 8x 8x will be slowed down a bit, then the memory on top is slowing the amd down a little bit too, plus the motherboard not being fx will slow it down too.

But it still is the same kind of fps at higher resolution isnt it? if the lowest resolution is removed they are all very close except for far cry and if it was an enthusiast motherboard used the amd would probably win most of them.
December 19, 2009 9:43:21 PM

jimmysmitty said:
Well its all software when its two GPUs in CG/SLI. With the x2 GPUs it turns more hardware based.

My only meaning is that its possible for both GPUs to be unhappy sharing one x16 link but probably not for a few more gens.


I provided evidence, you did not. HD5890 isn't PCIe bottlenecked, PERIOD. It's got nothing to do with this thread anyway so give it a rest.
This thread is about CPU-GPU scaling. The only reason why I bought up SLi/CrossFire was the extra CPU utilization issue e.g. HD4870X2 needs more CPU than HD5870 to scale.

@those looking at Xbitlab results
Look at minimum FPS for Crysis and WIC @1920x1200 for single HD5870. AFAIK both of those games are more GPU dependent than others, yet a higher clocked i7 9xx can still raise performance.

For single-GPU gaming platform, i5 750 is probably now a better choice than Phenom II now after seeing the above results.
a b à CPUs
December 20, 2009 10:21:24 AM

Ok lets have a closer look at the legionhardware article http://www.legionhardware.com/document.php?id=869

Starting with the facts.

i7 920 @ 2ghz - 4ghz
- x3 2GB G.Skill DDR3 PC3-12800 (CAS 9-9-9-24)
- Seagate 500GB 7200-RPM (Serial ATA300)
- ATI Radeon HD 5970 (2GB)
- ASUS P6T Deluxe (Intel X58)

Software
- Microsoft Windows 7 Ultimate (64-bit)
- ATI 8.663.1 Beta5 Hemlock Nov11

Phenom II 965 @ 2ghz - 4ghz
- x2 2GB G.Skill DDR3 PC3-12800 (CAS 9-9-9-24)
- Seagate 500GB 7200-RPM (Serial ATA300)
- ATI Radeon HD 5970 (2GB)
- ASUS M4A79T Deluxe (AMD 790FX)

Software
- Microsoft Windows 7 Ultimate (64-bit)
- ATI 8.663.1 Beta5 Hemlock Nov11


As you can see, both systems are as close to being identical as they can be so no excuses either way.

----------------------------------------------------------------------


Ok on to the benchmarks, starting with...

Batman AA -



With both cpu's at 2ghz, the i7 920 enjoys a 9.6% advantage. This gradually decreases all the way up to 4ghz, where the advantage is reduced to 6.3%. This will be a recurring theme.

Wolfenstein -



Generally so close that there is nothing in it, the Phenom II 'wins' by a single fps but results like that one make me think there is a hard cap on fps somewhere in the game.

Left4Dead is quite interesting.



At 2ghz the i7 has a 24% fps advantage - or 23fps. However this advantage is only 10 frames at 2.6ghz, or less than 10% already. At the top end of the clocks, the advantage is about 1-3 fps which is 1-2%. This is the second out of three benchmarks where we have seen the Phenom II reduce the fps at higher clock speeds.

CoD: MW2



At 2ghz the i7 has a commanding advantage of 33% or 29 fps. This advantage actually increases until 2.8ghz, at which point the Phenom II starts to close the gap quite remarkably. Just looking at the graph you can see the Phenom II is scaling like you would expect it to around it's regular bin speed.

In contrast, the i7 is going nowhere after its normal base ghz. This culminates in the Phenom II actually overtaking the i7 at 4ghz, resulting in a huge swing in favour of the AMD cpu from 2ghz - 4ghz.

At 2ghz the i7 won by 33%, at 4ghz the Phenom II won by a few (1-2%) percent.

Note :- This is the 3rd out of 4 benchmarks where we have seen a swing to the Phenom II at higher clock speeds.

Company of Heroes



At 2ghz the i7 has a commanding 33% lead again. @ 4ghz the lead is reduced to ~9%. Otherwise you can see some scaling on both cpu's.

Crysis Warhead



At 2ghz the i7 enjoys a 25% advantage, however this advantage is completely gone by 3.2ghz. In the end, both cpu's cannot push the game past 43 fps, which is what the phenom II manages at stock compared to the i7 with a 50% overclock. (note i just added that last sentence to show how easily you can twist benchmarks to your own agenda).

Far Cry 2



A single frame i7 lead at 2ghz turns into a 3 fps lead for the phenom II quite quickly. You want to talk percentages? The Phenom II is ~5% faster overall.

Battleforge



This looks like a cookie cutter of most of the others. An (46fps vs 39fps) i7 18% lead at low clock speeds turns around to a small Phenom II lead at normal - max clocks.

Hawx



@2ghz the i7 starts off with it's usual advantage. In this case it's only around 5% overall though. That quickly dissipates and the Phenom II takes over at around the i7's normal clock speed (~2.6ghz). As the clock speed increases, the Phenom II starts to pull ahead so that at 3.8ghz, it enjoys a 3% lead).

-------------

How many times in the above graphs did you see the i7 start off well with a big lead at low ghz, and fall away at higher ghz? It looks a bit similar to the i7 performing well at low resolution and falling away at higher resolution doesnt it?

This is apples to apples, as close as it can be - just look at the test setups.

All I want to add to this right now is - ALL of these games were designed on intel cpu's.

The Phenom II generally keeps increasing, or scaling like you would imagine it would happen while the i7's hit a brick wall and stop scaling.

In many of the games, a higher clocked Phenom II's would continue to increase the fps while higher clocked i7's would go nowhere.

The most obvious thing we can see is, overclocking the i7 gives extremely disappointing scaling most of the time.

Regardless, there is no doubt that the best 'single card' gaming setup available right now is a 5970 + Phenom II 965 BE and regardless of what elmo claims, this proves without any doubt that the Phenom II is faster at higher clocks. Even faster clocked Phenom II's will increase this gaming advantage over the i7 while higher binned i7's will not do anything to close the gap.

I know what you are thinking - 'but the 5870 xfire results on xbit prove the i7 is faster'. Well no they don't and I'll explain that too.

1) As jamahl mentioned, this was never a fair test and to xbit's credit at least they managed to mention that too.

2) Two 5870's are actually a lot different to a 5970. The 5970 is internally crossfired and just works while two 5870's will need driver optimisations. Think back to a few months ago when the i7 had a commanding lead with crossfire until ATI released the 9.8 driver which pulled the phenom II right back to almost equal (remember Farcry 2? There is a reason why the i7 has a huge lead in crossfire with that again using 5xxx cards).

There will be another crossfire driver doing the same thing for the 5xxx's soon and that farcry lead will be gone, and elmo will be looking for another excuse. I 'guarantee' it.
a b à CPUs
December 20, 2009 10:31:24 AM

Just to add - of course the i7 is going to be much faster at 2ghz. Think about it.

The i7 start at 2.66 ghz compared to the Phenom II 965's 3.4ghz. The Phenom II has to downclock massively to get to 2ghz in comparison.

However the same is true going the other way. The i7 shows diminishing returns at much higher clocks while the Phenom II is quite comfortably gaining fps continuously as the clocks go higher.

If both chips could go to 5ghz and the scaling continued, the Phenom II would probably have a 10% fps lead in most games. Unfortunately it doesn't so it's unlikely that we will ever know for sure.
a b à CPUs
December 20, 2009 10:59:27 AM

Wake up from dreamland. The 965 doesn't just generally keep increasing. That basically the same as the clock speeds go up. The difference as you clearly point out is the i7 is faster at lower clocks. So it takes lower Ghz to get rid of any cpu bottleneck. While the AMD is slower clock for clock so it "scales" better on the charts because it is starting at such a low clock that it really hurts the fps. Alot more than the Intel.

The AMD gains very little going up to 4ghz from its default 3.4.

Here are the gains.

Batman 155 to 159.

Wolf 92 to 93

Left 127 to 130

Call 124 to 132

Company 100 to 102

Crysis 42 to 43

Far cry. 42 to 42. No change.

Battle 46 to 47

Tom Clancys 85 to 86


The Phenom is not at all quite comfortable gaining fps as the clocks go higher.

When AMD hits the same FPS wall the Intel does. The AMD gets a 1-3 FPS lead when this happens. Which is what Elmo has been saying.


If the AMD could go to 5ghz the scaling would not continue.

You are such a blinded AMD fan it is rather sad.

a b à CPUs
December 20, 2009 11:07:03 AM

Let's look at it logically shall we?

Crysis Warhead, Farcry 2 and Battleforge - the only games that do not run perfectly at maximum settings.

Crysis Warhead is a draw, Farcry 2 a win for the Phenom II, Battleforge a win for the Phenom II.

The Phenom II is the best single card gaming cpu choice. EOD.
a b à CPUs
December 20, 2009 11:12:54 AM

Btw the Phenom II gains 21 fps from 3.4ghz to 4ghz while the i7 only gains 14. Thats what, 33% better at high resolution high clocks? *chuckles*

Yes you can twist benchmarks to say what they like, but you cannot change the facts either. The Phenom II is the best current gaming cpu - at least until we see the i5 results.
a b à CPUs
December 20, 2009 11:20:22 AM

The *tougher* the game is to run, the better the Phenom II does. Just look at it - how hard is it to see? When the fps is in the 100's, (Batman, CoH, L4d) the i7 does better except for at the very end in MW2.

When the system is really struggling like in Crysis, Farcry 2 and Battleforge - the Phenom can squeeze out another couple of frames where it matters most.

Look at those tables and argue me that the Phenom II is not the superior gaming choice.
a b à CPUs
December 20, 2009 11:28:48 AM

That is fine if thats how you want to put it now after that entire scaling thesis failed. When there is a GPU bottleneck the AMD chip shows it 1-3 FPS faster. Again it looks like its because of what Elmo has been suggesting with the QPI/PCI. Regardless of what causes it, it is there.

But also keep in mind the cpus running at stock settings. At "stock" settings the Intel is either dead even with the 965 or 1-3fps ahead. I put the quotes up because the Intel is not using turbo.

This review debunks basically of all BS talking points the AMD delusional folks love to repeat.

What it does show again is when the gpu is the bottleneck the AMD is 1-3 fps faster. But then can never be enough. They always got to come up with some BS spin.



a b à CPUs
December 20, 2009 11:36:16 AM

jennyh said:
Btw the Phenom II gains 21 fps from 3.4ghz to 4ghz while the i7 only gains 14. Thats what, 33% better at high resolution high clocks? *chuckles*

Yes you can twist benchmarks to say what they like, but you cannot change the facts either. The Phenom II is the best current gaming cpu - at least until we see the i5 results.



Are you kidding me. That is you BS spin of the moment. Lets just add up the total fps? I dont see any game in there that gains 21 so I assume that is what you are doing. *chuckles at the latest installment of stupidity*
a b à CPUs
December 20, 2009 11:47:16 AM

someguy7 said:
Are you kidding me. That is you BS spin of the moment. Lets just add up the total fps? I dont see any game in there that gains 21 so I assume that is what you are doing. *chuckles at the latest installment of stupidity*


The fact is, from 3.4ghz to 4ghz, the Phenom II increases framerate by 33% over that whole choice of games.

You see how easy it is? Why is it that when Elmo puts his spin on things you aren't jumping all over that?

The *only* fact that matters is that the Phenom II is superior when it needs to be - that is below 60fps. You can say that the Phenom II copes better with bottlenecks or use whatever fancy language you like but the end result is a win for the Phenom II.

This self-styled 'seeker of knowledge and truth' elmo doesn't seem to want to point that out does he? I wonder why. :whistle: 
a b à CPUs
December 20, 2009 12:45:53 PM

That is no fact at all. I wonder how you even came up with that number actually. That statement is actually comical. Are you even aware of how to a get a percentage of framerate increase? None the less. I will not spend my time giving you a math lesson.

So know I don't see how easy it is.

As far as Elmo goes. I stated in this very thread that the 2ghz performance is pointless but its perfect for the intent of the article. Elmo is not in here spinning.

There are 3 games in this that never crack 60fps. At stock settings(no turbo) two are even and the other one Intel wins by ONE frame. At 4ghz one is even. One has AMD ahead by one frame. The last one is 3frames ahead.

There is no fancy language. Its truth. It would make perfect sense that you didn't actually understand the language though.

What doesn't Elmo want to point out exactly?
a b à CPUs
December 20, 2009 1:11:06 PM

elmo said...

Quote:
These users mistakenly assume this is some sort of proof that "AMD is the better gaming chip offering better performance" as one user suggested. This assertion is patently false.


It is not patently false, it is a simple fact and anyone who knows anything about gaming will agree that 3 fps below 60fps is worth infinitely more than 10 fps above 100fps.

This is why the Phenom II *is* the superior gaming chip, and that is what elmo *isn't* pointing out.
a b à CPUs
December 20, 2009 1:27:55 PM

Now lets add in a second card... or switch out for 2 5870s...

Oh damn Phenom II loses...

Now lets take the sums of all this data and give it a non-biased scew:

Phenom II performs the same as Core i5/i7 with a single GPU. Hasn't that been said for months?

And no the scaling difference does not show up with a single 5970.
a b à CPUs
December 20, 2009 1:33:33 PM

You realise that the 5970 is a dual gpu card right?

The reason the xbitlabs article showed the phenom II behind is due to 1) drivers and 2) a lower spec motherboard and ram combo.

This has been proven plenty in the past. There is no huge increase for the i7 with dual-gpu's that hasn't been fixed with drivers.

Quite why the i7 performs better 'out of the box' is anyone's guess, but I'd imagine it might have something to do with every game being designed on an i7? Therefore, AMD can only catch up with driver revisions a couple of months later.

THE CPU IS JUST AS GOOD AT GAMING - single or crossfire.

How exactly does elmo's theory work out in terms of crossfire btw? If the i7 is bottlenecking a single card before the Phenom II does, how exactly does it turn that around with dual gpu's?

Quote:
Well I think I've found the answer and have tried to painstakingly spread the answer in order to further knowledge. Intel holds a patent which suggest (clearly states actually) that there is an inherent incompatibility between the communications protocol the PCI Express bus uses and that of the QPi link. This patent explains how the IOH (X58 chip) acts as a hub converter and "tunnels" the data through (in one cycle). This "cycle" means latency in lay mans terms.


It can't cope with one gpu but it can cope better with two? Something fails here logically no?

The real answer is - it doesn't, the i7 is faster *currently* until ATI get the drivers sorted out. Just like the 9.8 catalyst drivers sorted out the crossfire issues with the 4-series cards, there will soon be drivers sorting out the issues with 5-series.
a b à CPUs
December 20, 2009 1:58:46 PM

No you don't get it jenny. First of all the CPU matters more for 2 5870s than 1 5970 but that wasn't my point. My point was that when you add more GPU horsepower and the GPU bottleneck is gone then the barrier that the Core i7 920 is hitting will be gone and it will almost certainly outperform the Phenom II. That is the point trying to be made.

What is missing from that point, at least in this thread, is thus:

1. Core i5/i7 will be faster, yes, but not by a very large amount. The Core i5/i7 will never be tangibly faster in gaming, sure the benchmarks will love it, but the actual gaming experience will remain the same. This should run parallel with fully quadcore optimized games, for some reason I have seen people think that the Core i5/i7 performs better in quadcore optimized games than Phenom, but the difference will be the same as any other game since they are both quadcores.

2. Phenom II can now fully compete with Core i5/i7 when there are multiple ATI cards in use thanks to drivers and the symbiotic bond of the parts in the platform. This is not true for multiple nVidia cards. There is no driver that can simulate this on an AMD board because there are no hardware compatibilities, even though the boards are made by nVidia in this case, they do not work as well with the AMD CPU. This means that Core i5/i7 will match Phenom II with multiple ATI cards and will pull ahead of Phenom II with multiple nVidia cards.
a b à CPUs
December 20, 2009 2:09:00 PM

If there weren't already a bunch of benchmarks showing the Phenom II running better on nvidia cards I might actually agree with that.

You are completely wrong about the i7 breaking down the barriers with faster gpu's. It didn't do it with the 4870 or g200's at max settings, it isn't doing it with the 5970.

It wont do it with the 6970 or whatever it is called. At maximum settings - the settings one would expect to use with top end equipment - the i7 will always fall behind the Phenom II or match it at the best.

The benchmarks above are the proof of it - the i7 cannot cope as well with seriously demanding games, it just makes it look better on benchmarks doing 'easy' games. The next demanding games on maximum settings will *still* be better on the Phenom II at max settings. Now if you want to talk about dropping settings etc to make the i7 run better that's fine - but the proof clearly shows that the Phenom II copes better with more demanding games and max settings.
a b à CPUs
December 20, 2009 2:29:08 PM

Ok I quit on this one, maybe someone else has the time to explain to you why you are wrong.
a b à CPUs
December 20, 2009 2:30:21 PM

No please you explain it to me.

I've been listening to intel fanboys saying the same thing all year btw.

You would see a benchmark or two that showed the i7 scoring 10-20fps more at fps totals near the 130 mark and make the assumption that it was better at gaming.

A year ago an i7 running FarCry 2 on medium settings would indeed have scored 10-20 more fps than a Phenom II. However if you ran Crysis on max they were both about the same (19-20 fps terrible).

People said that when the next generation of cards came out, the i7 would pull further ahead based on the FarCry 2 results.

The cards are out and that didn't happen. What happened was, at maximum settings and a faster graphics card (5970), Far Cry 2 runs better on the Phenom II.

Give me one good reason why a 6970 would run games on max better on an i7 than it would a Phenom II.
December 20, 2009 2:58:36 PM

Id throw out the xbitlabs benches as theyve often shown very unreliable benches in many things, not just cpus, but also power tests etc, many things, gpus etc, so not a good example, just sayin.
Its known SMT causes slow downs vs non SMT, and this may be what were seeing, as a x2 solution will be causing mor work for the cpu.
Wed really need to see games thatre demanding and well MT'd, and then go from there.
Id never expect the i7 to ever be much more than 10% better, as its IPC is around that of the P2 and C2Q, anything higher always shows SMT benefits which again, if not working properly causes slowdowns per clock, not better
a b à CPUs
December 20, 2009 3:10:36 PM

Jaydee you know better than most what we've been reading the past year.

The i7 was far superior in crossfire for a while - 'proof' of it's superiority. A simple driver update changed all that.

The i7 was supposed to pull further ahead with overclocking - That legionhardware article shows the Phenom II actually closes the gap and sometimes pulls ahead with overclocking.

The i7 would perform better with more powerful gpu's - proven wrong again.

It is *exactly* like I said it would be months ago - the i7 would continue to 'win' when the fps totals are above 100fps, but would continue to struggle at maximum settings that stressed the system.

That legionhardware article completely vindicates everything I said would happen - now I'm pushing it on again and saying that in a years time or so, a 6970 will show the exact same situation as we are seeing with the 5970 right now and the Phenom II will still be beating the i7 at max-settings.
December 20, 2009 3:29:15 PM

i cannot speak for the 5970 platform vs platform but in my own little world, my 295 plays much better (single 295) on my 955 then it did on my 920. smoother and cleaner but i am one guy, one experience but i figured i would throw that on the pile.
December 20, 2009 3:39:18 PM

My old thread regarding all this had links from AMD saying the new MT drivers would show better gains on some cpus than others, and some were already shouting unfair etc, as if AMD wouldnt allow for better fps accross the board, but actually, it only allowed for better access/thruput from cpus which didnt already have it.
Hats off to Intel here for their solution thats been fixed with SW, as their cpus were able to handle it without it, but again, Intels also worked hand in hand with M$ to make this work, so the GFX companies, nVidia included, as we saw on the older Anand benches which also needed SW solutions from them at the time, needed to correct this, as its a communication problems thru SW, and not a HW solution, and yes, I know all the claims, and its simply not there.
The cpu has to work thru the drivers, if those drivers arent focused on whats available on the cpu, then we will see these lower numbers, but when done right, itll show what it should, no more than 10%, and the actual differences in each cpus arch, and how they respond to the demands put upon them, as we see fluctuations in differing apps, so to we see fluctuations with differing games, as each arch works differently with each game, and id point out, as many have here, using games to determine a cpus ability isnt the best way to do it, only using the same game per evidence, but using games overall for games can be done, excluding other apps, but more on a game to game scenario, and somewhat also on todays harder more demanding games in a more general sense, tho that too has many exceptions
a b à CPUs
December 20, 2009 9:18:15 PM

Alright Jenny here is why you are wrong:

1. The point about CFX vs SLI performance is true, the drivers that increased CFX performance did so because they allowed the AMD CPU to work with the ATI GPUs more efficiently. That driver was an ATI card driver, the same was never released for nVidia because the nVidia cards were not built around that.

2. You consistently say that Phenom II beats Core i5/i7 and that Core i5/i7 "struggles" at high end gaming. This is not the case, the Phenom II has really brought the hurt on the Core i5/i7 by being able to match or come close to Core i5/i7 in gaming. This is a great accomplishment and really makes the Phenom II a great cost efficient system that can easily manage high end gaming like any other CPU on the market now.

3. You seem to think that games are magical, they are not. They are applications. If a game comes along that uses 100% of both CPUs then the Core i5/i7 will be faster than Phenom II like in every other instance of 100% usage. Games are not magical. That said, games will come along that do that, but they will not NEED that much CPU power in comparison to GPU power. This means that Core i5/i7 will likely be only slightly faster in this instance, perhaps the difference will even be negligible like under a FPS difference.

4. Lastly, you take a few benchmarks and treat them like fact. You take a little difference and make that into all the difference in the world. If you think you can see an average of 3FPS when over 30 average FPS then you are very mistaken. If you think that ANY comparison is accurate enough to make a difference of 3 FPS a sure difference then you are also very mistaken.

The point is of all this, something you nor Elmo seem to see, is that in gaming with multiple ATI cards, a Core i5/i7 will show no tangible benefit over a Phenom II. When nVidia cards are in the equation there might be a slight difference, but no one should buy multiple nVidia cards to put in an AMD system because nVidia boards are pure ***, so that is irrelevent. If AMD manages to allow SLI on their boards like Intel does, then a driver update will likely come to add the same improvements.

Regardless, pure gamers should stick with Phenom II, or Core i5 if they need the SLI support.

For non-pure gamers, then that is a different story and is very dependent on budget and the needs of the individual.
a c 123 à CPUs
December 20, 2009 9:44:26 PM

wuzy said:
I provided evidence, you did not. HD5890 isn't PCIe bottlenecked, PERIOD. It's got nothing to do with this thread anyway so give it a rest.
This thread is about CPU-GPU scaling. The only reason why I bought up SLi/CrossFire was the extra CPU utilization issue e.g. HD4870X2 needs more CPU than HD5870 to scale.

@those looking at Xbitlab results
Look at minimum FPS for Crysis and WIC @1920x1200 for single HD5870. AFAIK both of those games are more GPU dependent than others, yet a higher clocked i7 9xx can still raise performance.

For single-GPU gaming platform, i5 750 is probably now a better choice than Phenom II now after seeing the above results.


Calm down man. My point is that since the 5970 is internally CF it runs differently. Sometimes its faster than 2 5870s others its not.

roofus said:
i cannot speak for the 5970 platform vs platform but in my own little world, my 295 plays much better (single 295) on my 955 then it did on my 920. smoother and cleaner but i am one guy, one experience but i figured i would throw that on the pile.


Wait... Phenom 955 vs a Core i7 920? Depends on the games too though. Some still take more advantage of a higher clock, hence why a dual core C2D can keep up in those games while others rely on cores (few) and lastly, some rely on pure IPC.

JAYDEEJOHN said:
My old thread regarding all this had links from AMD saying the new MT drivers would show better gains on some cpus than others, and some were already shouting unfair etc, as if AMD wouldnt allow for better fps accross the board, but actually, it only allowed for better access/thruput from cpus which didnt already have it.
Hats off to Intel here for their solution thats been fixed with SW, as their cpus were able to handle it without it, but again, Intels also worked hand in hand with M$ to make this work, so the GFX companies, nVidia included, as we saw on the older Anand benches which also needed SW solutions from them at the time, needed to correct this, as its a communication problems thru SW, and not a HW solution, and yes, I know all the claims, and its simply not there.
The cpu has to work thru the drivers, if those drivers arent focused on whats available on the cpu, then we will see these lower numbers, but when done right, itll show what it should, no more than 10%, and the actual differences in each cpus arch, and how they respond to the demands put upon them, as we see fluctuations in differing apps, so to we see fluctuations with differing games, as each arch works differently with each game, and id point out, as many have here, using games to determine a cpus ability isnt the best way to do it, only using the same game per evidence, but using games overall for games can be done, excluding other apps, but more on a game to game scenario, and somewhat also on todays harder more demanding games in a more general sense, tho that too has many exceptions


You mean how Intel released info and said that it would use its CPU to offload some work from the GPU was called unfair?

Funny how that works. Two faced kinda. I even pointed that out. AMD made drivers that give a performance boost for Dragon based systems. It wasn't called unfair. Intel does the same it is.

Oh well.

Anyone else notice how this thread went down hill recently?
a b à CPUs
December 20, 2009 9:46:07 PM

How can a fundamentally wrong thread go downhill exactly?
a c 123 à CPUs
December 20, 2009 9:51:04 PM

When someone who disagrees with pretty much every mass agreement says its wrong all the time.

It could be clear as day to you and you would say it was wrong just like those who said the games were all rigged in Intels favor during the Core 2 days.

Sad....
a b à CPUs
December 20, 2009 9:56:23 PM

Well to be honest the beginning point, while not necessarily "wrong", was awfully over drawn and biased. If the conditions come around where there is a CPU bottleneck then, yes, the Core i5/i7 will do better than the Phenom II, but its not going to make a huge difference since the GPU is still the most driving component.
a b à CPUs
December 20, 2009 9:57:12 PM

Mass agreement of intel fanboys is simply a mass agreement of intel fanboys.

I notice elmo has been pretty quiet since the start? Maybe he's finally learned when to give up a loss, the rest of you need to do the same.

And no offence Jimmy but I told you intel were crooked, i told you AMD's share price would rocket, I told you Cypress would be out 6 months before Fermi, I told you Larrabee was dead and I told you that the Phenom II would remain top gaming cpu at max resolution...all against a myriad of intel fanboys who disagreed.

If you even get one thing right? Maybe I'd listen.
a b à CPUs
December 20, 2009 10:00:19 PM

jennyh said:
Mass agreement with intel fanboys is simply a mass agreement of intel fanboys.

I notice elmo has been pretty quiet since the start? Maybe he's finally learned when to give up a loss, the rest of you need to do the same.


I'm pretty sure there were a lot of mass disagreements in this thread, but because you AND ONLY YOU see it differently than everyone else, that makes everyone else wrong, biased, and fanboys.

Though I agree Elmo was trying for the Pro-Intel sensationalism... which ended up into a flame war thread.
a b à CPUs
December 20, 2009 10:11:34 PM

It's not a flame thread, you need to understand my style. When 5-10 people argue against you, you need to have a certain style else you'll get eaten alive.

I will not be browbeaten by the flawed majority. It's actually very simple for me because I have an unshakeable faith in the knowledge that what I am doing is right.

And they are wrong, just like they were wrong on Larrabee, wrong on intel's crimes, wrong on AMD going bust and everything else.
December 20, 2009 10:42:16 PM

jimmysmitty said:
Calm down man. My point is that since the 5970 is internally CF it runs differently. Sometimes its faster than 2 5870s others its not.



Wait... Phenom 955 vs a Core i7 920? Depends on the games too though. Some still take more advantage of a higher clock, hence why a dual core C2D can keep up in those games while others rely on cores (few) and lastly, some rely on pure IPC.



You mean how Intel released info and said that it would use its CPU to offload some work from the GPU was called unfair?

Funny how that works. Two faced kinda. I even pointed that out. AMD made drivers that give a performance boost for Dragon based systems. It wasn't called unfair. Intel does the same it is.

Oh well.

Anyone else notice how this thread went down hill recently?


No, and youre saying the i7 is twice as fast at everything at least than any previous cpu?
I can take anything you say out of context as well, and waste my time answering you, but I wont bother.
What Ive said has absolutly nothing to do with what youre refering to, but it sure makes me look bad taking me out of context this way, good going, just keep those 2x anythings comments a comin
a c 123 à CPUs
December 21, 2009 3:45:19 AM

jennyh said:
It's not a flame thread, you need to understand my style. When 5-10 people argue against you, you need to have a certain style else you'll get eaten alive.

I will not be browbeaten by the flawed majority. It's actually very simple for me because I have an unshakeable faith in the knowledge that what I am doing is right.

And they are wrong, just like they were wrong on Larrabee, wrong on intel's crimes, wrong on AMD going bust and everything else.


And thats why you will never get anywhere. If you think you are always right, you have a long way to go. No one is ever always right. Even Stephen Hawkings is never always right.

Back a long time ago, people always though AMD would just be a Intel clone forever. Those people stood firm in their "knowledge". Just a bit ago some people thought Phenom would kill C2Q in everything. They also were firm in their "knowledge".

Not sure why I would watse time with you oh so smart and right about everything.

JAYDEEJOHN said:
No, and youre saying the i7 is twice as fast at everything at least than any previous cpu?
I can take anything you say out of context as well, and waste my time answering you, but I wont bother.
What Ive said has absolutly nothing to do with what youre refering to, but it sure makes me look bad taking me out of context this way, good going, just keep those 2x anythings comments a comin


JDJ, do you just have it out for everyone whos against your ideals? I point out something and you get really defensive about it. You said people were claiming AMDs performance boost was unfair. You said it wasn't. I pointed that some people decided to say Intel doing the same thing was unfair while also saying AMD doing it was fair.

BTW, Core i7 is able to give 2x the performance in a lot of apps, especially ones that need memory bandwidth. But I forget. Outside the realm of gaming, it doesn't ever matter.....
December 21, 2009 6:31:15 AM

Now your onto something completely different jimmy.
What youre talking about now was my comments on those who commented on AMDs MT boost with their drivers.
Those same people acted like AMD was making drivers that work only on AMD cpus, where I tried to explain that wasnt the case, and Ive already described it here, some cpus get more benefits from them than others, as some already had those benefits.

I dont think thatd be right if AMD were to do this, and frankly they, like I, see it as a dead end, only hurting their sales and customers in the end, and is a foolish tact, but none the less, there were people here who thought as much, that theyd do such a thing.


OK, I guess I will have to address this, The cpu offloading was in reference to Intels IGPs, and that only, from another thread, and to claim Intels IGPs in only a few games where this was happening as them being better when all the work was done on the cpu is, well I dunno, foolish? Disingenuine? Flat out lying? Who knows? Its not the IGP doing the work, and it isnt in every game, and where it isnt applied, its a stinky Intel IGP heheh.
So, maybe now you can see how I saw you takingthis out of context? Mixing Intels real actions, like their bribes and cpu/IGP usage, vs AMDs improvements and their new MT drivers which helped all cpus, just some more than others, like Intel illegal activities and effects arent so compatable.
While Intel did do the cpu helping the IGP, and played it up, and only on a few games at that, AMD made a driver for all solutions, with varying effects on effectiveness
a b à CPUs
December 21, 2009 6:47:37 AM

jennyh said:
Mass agreement of intel fanboys is simply a mass agreement of intel fanboys.

I notice elmo has been pretty quiet since the start? Maybe he's finally learned when to give up a loss, the rest of you need to do the same.

And no offence Jimmy but I told you intel were crooked, i told you AMD's share price would rocket, I told you Cypress would be out 6 months before Fermi, I told you Larrabee was dead and I told you that the Phenom II would remain top gaming cpu at max resolution...all against a myriad of intel fanboys who disagreed.

If you even get one thing right? Maybe I'd listen.


jennyh said:
Ok lets have a closer look at the legionhardware article http://www.legionhardware.com/document.php?id=869

Starting with the facts.

i7 920 @ 2ghz - 4ghz
- x3 2GB G.Skill DDR3 PC3-12800 (CAS 9-9-9-24)
- Seagate 500GB 7200-RPM (Serial ATA300)
- ATI Radeon HD 5970 (2GB)
- ASUS P6T Deluxe (Intel X58)

Software
- Microsoft Windows 7 Ultimate (64-bit)
- ATI 8.663.1 Beta5 Hemlock Nov11

Phenom II 965 @ 2ghz - 4ghz
- x2 2GB G.Skill DDR3 PC3-12800 (CAS 9-9-9-24)
- Seagate 500GB 7200-RPM (Serial ATA300)
- ATI Radeon HD 5970 (2GB)
- ASUS M4A79T Deluxe (AMD 790FX)

Software
- Microsoft Windows 7 Ultimate (64-bit)
- ATI 8.663.1 Beta5 Hemlock Nov11


As you can see, both systems are as close to being identical as they can be so no excuses either way.

----------------------------------------------------------------------


Ok on to the benchmarks, starting with...

Batman AA -

http://i293.photobucket.com/albums/mm62/Xcpus_Tactician/Batman_Combined.png

With both cpu's at 2ghz, the i7 920 enjoys a 9.6% advantage. This gradually decreases all the way up to 4ghz, where the advantage is reduced to 6.3%. This will be a recurring theme.

Wolfenstein -

http://i293.photobucket.com/albums/mm62/Xcpus_Tactician/Wolfenstein_Combined.png

Generally so close that there is nothing in it, the Phenom II 'wins' by a single fps but results like that one make me think there is a hard cap on fps somewhere in the game.

Left4Dead is quite interesting.

http://i293.photobucket.com/albums/mm62/Xcpus_Tactician/L4D2_Combined.png

At 2ghz the i7 has a 24% fps advantage - or 23fps. However this advantage is only 10 frames at 2.6ghz, or less than 10% already. At the top end of the clocks, the advantage is about 1-3 fps which is 1-2%. This is the second out of three benchmarks where we have seen the Phenom II reduce the fps at higher clock speeds.

CoD: MW2

http://i293.photobucket.com/albums/mm62/Xcpus_Tactician/COD-MW2_Combined.png

At 2ghz the i7 has a commanding advantage of 33% or 29 fps. This advantage actually increases until 2.8ghz, at which point the Phenom II starts to close the gap quite remarkably. Just looking at the graph you can see the Phenom II is scaling like you would expect it to around it's regular bin speed.

In contrast, the i7 is going nowhere after its normal base ghz. This culminates in the Phenom II actually overtaking the i7 at 4ghz, resulting in a huge swing in favour of the AMD cpu from 2ghz - 4ghz.

At 2ghz the i7 won by 33%, at 4ghz the Phenom II won by a few (1-2%) percent.

Note :- This is the 3rd out of 4 benchmarks where we have seen a swing to the Phenom II at higher clock speeds.

Company of Heroes

http://i293.photobucket.com/albums/mm62/Xcpus_Tactician/CompanyofHeroes_Combined.png

At 2ghz the i7 has a commanding 33% lead again. @ 4ghz the lead is reduced to ~9%. Otherwise you can see some scaling on both cpu's.

Crysis Warhead

http://i293.photobucket.com/albums/mm62/Xcpus_Tactician/CrysisWarhead_Combined.png

At 2ghz the i7 enjoys a 25% advantage, however this advantage is completely gone by 3.2ghz. In the end, both cpu's cannot push the game past 43 fps, which is what the phenom II manages at stock compared to the i7 with a 50% overclock. (note i just added that last sentence to show how easily you can twist benchmarks to your own agenda).

Far Cry 2

http://i293.photobucket.com/albums/mm62/Xcpus_Tactician/FarCry2_Combined.png

A single frame i7 lead at 2ghz turns into a 3 fps lead for the phenom II quite quickly. You want to talk percentages? The Phenom II is ~5% faster overall.

Battleforge

http://i293.photobucket.com/albums/mm62/Xcpus_Tactician/BattleForge_Combined.png

This looks like a cookie cutter of most of the others. An (46fps vs 39fps) i7 18% lead at low clock speeds turns around to a small Phenom II lead at normal - max clocks.

Hawx

http://i293.photobucket.com/albums/mm62/Xcpus_Tactician/HAWX_Combined.png

@2ghz the i7 starts off with it's usual advantage. In this case it's only around 5% overall though. That quickly dissipates and the Phenom II takes over at around the i7's normal clock speed (~2.6ghz). As the clock speed increases, the Phenom II starts to pull ahead so that at 3.8ghz, it enjoys a 3% lead).

-------------

How many times in the above graphs did you see the i7 start off well with a big lead at low ghz, and fall away at higher ghz? It looks a bit similar to the i7 performing well at low resolution and falling away at higher resolution doesnt it?

This is apples to apples, as close as it can be - just look at the test setups.

All I want to add to this right now is - ALL of these games were designed on intel cpu's.

The Phenom II generally keeps increasing, or scaling like you would imagine it would happen while the i7's hit a brick wall and stop scaling.

In many of the games, a higher clocked Phenom II's would continue to increase the fps while higher clocked i7's would go nowhere.

The most obvious thing we can see is, overclocking the i7 gives extremely disappointing scaling most of the time.

Regardless, there is no doubt that the best 'single card' gaming setup available right now is a 5970 + Phenom II 965 BE and regardless of what elmo claims, this proves without any doubt that the Phenom II is faster at higher clocks. Even faster clocked Phenom II's will increase this gaming advantage over the i7 while higher binned i7's will not do anything to close the gap.

I know what you are thinking - 'but the 5870 xfire results on xbit prove the i7 is faster'. Well no they don't and I'll explain that too.

1) As jamahl mentioned, this was never a fair test and to xbit's credit at least they managed to mention that too.

2) Two 5870's are actually a lot different to a 5970. The 5970 is internally crossfired and just works while two 5870's will need driver optimisations. Think back to a few months ago when the i7 had a commanding lead with crossfire until ATI released the 9.8 driver which pulled the phenom II right back to almost equal (remember Farcry 2? There is a reason why the i7 has a huge lead in crossfire with that again using 5xxx cards).

There will be another crossfire driver doing the same thing for the 5xxx's soon and that farcry lead will be gone, and elmo will be looking for another excuse. I 'guarantee' it.


You know... generally I have the patience of an angel. I could go one and refute you point by point using sound reasoning and logic. I have gone to great lengths in trying to instruct you. I am the teacher and you are the student. You somehow think it's the other way around. I've attempted to reason with you and I think that is where I've gone wrong. So let's try this method...


You are, quite possibly, the biggest idiot on the INTERNET. You're worse than VenomFangX, Scentia and Abinstein put together because unlike them you cannot claim that we never tried to instruct you. YOU CHOOSE TO IGNORE EVERYTHING WE SAY. There is no other way to describe the idiotic statements coming out of your mouth. How the hell is a diminishing return in Frames Per Second in a GAMING BENCHMARK as you scale up the CPU clocks indication of anything other than A GPU BOTTLENECK! (taking into consideration the rest of the information I've shared with the Xbit tests).

Holy CRAP you're ignorant. LISTEN TO US WE ACTUALLY WORK IN THE INDUSTRY!

Seriously.
a b à CPUs
December 21, 2009 7:49:29 AM

Elmo...you failed to understand that I was simply putting another spin on the benchmarks that you attempted to spin first.

Now you've lost this one, would you rather talk about Larrabee again, or maybe how many tflops the 5870 can kick out?





No I didnt think so. Next moronic thread you post? I'll be ripping it to shreds again because you are wrong in every single thread you make and the reason is simply - you are blinded by the i7 and the biggest intel fanboy I have seen on this forum so far regardless of what you claim otherwise.

WRONG on Larrabee WRONG on the tflops of a 5870 WRONG on the i7 being a better gaming cpu and your head is so far up your own arse you cannot even admit it.
!