Sign in with
Sign up | Sign in
Your question

Hexus.net benchmarks Nehalem - Page 2

Last response: in CPUs
Share
August 19, 2008 2:55:37 PM

Unless these numbers change, I agree, the "review" is oversensationalized. It looks like a bump for a Penryn so far. Which is nice, but all Ive heard for awhile now is Nehalem. Im still hoping these numbers ARE off, cause from a gamers perspective, we NEED it to
August 19, 2008 3:04:44 PM

The same analysis Hexus did, but this time on Anand, much better and realistic.

Quote:

As IDF has started, the first benchmarks of Nehalem will probably pop up. It is without a doubt an impressive architecture that gets a much better platform to run on, but this CPU is not about giving you better frames per second in your favorite game than the Penryn family. Let me make that more clear: even when the GPU is not the bottleneck, it is likely that most games will not significantly faster than on Penryn. We, the people behind it.anandtech.com will probably have the most fun with it, more than your favorite review crew at Anandtech.com :-). And no, I have not seen any tests before I type this. Nehalem is about improving HPC, Database and virtualization performance, much less about gaming performance. Maybe this will change once games get some heavy physics threads, but not right away.

Why? Most Games are about fast caches and super integer performance. After all, most of the Floating point action is already happening on the GPU. All Core 2 CPUs were a huge step forward in integer performance (not in the least because of memory disambiguation) compared to the CPUs of that time (P4 and K8). Nehalem is only a small step forward in integer performance. And the gains due to slightly increased integer performance are mostly negated by the new cache system. In a previous post I told you that most games really like the huge L2 of the Core family. With Nehalem they are getting a 32 KB L1 with a 4 cycle latency, next a very small (compared to the older Intel CPUs) 256 KB L2 cache with 12 cycle latency and after that a pretty slow 40 cycle 8 MB L3. When running on Penryn, they used to get a 3 cycle L1 and a 14 cycle 6144 KB L2. That is a 24 times larger L2 than Nehalem!

The percentage of L2 caches misses of most games running on a Penryn CPU is extremely low. Now that is going to change. The integrated memory controller of Nehalem can't help much, as the fact remains that the L3 is slow and the L2 is small.

But that doesn't mean Intel made a bad choice. Intel made a superbly good choice by improving the performance where Core (Merom/Penryn) was mediocre to good. Penryn was already a magnificent gaming CPU, but it could not beat the AMD competitor in HPC benchmarks. And AMD gave good resistance in the database performance benchmarks. That is all going to change.

Most Database code can not use the wide architecture of Penryn very well. The number of instructions per cycle get lower than 0.5 and waiting for the memory is the most probably cause. SMT or Hyperthreading can do wonders here: while one thread waits for a memory stall, the other thread continues working and vice versa.

Secondly, quad (and eight) socket performance is going to improve a lot as four Nehalems only have to keep four L3 in sync, while a similar Tigerton system has to keep 8 L2 caches in sync. That is why the cache system is perfect for server performance, but a little less interesting for gaming performance.

The massive bandwidth that the integrated tri-channel memory controller delivers will do wonders for HPC code. And the new TLB architecture with EPT will make Nehalem shine compared to it's older Core brothers.

No, Nehalem was made to please the IT and HPC people. Bring it to it.anandtech.com, it is not that interesting for you gamers ;-)



Link: http://www.anandtech.com/weblog/showpost.aspx?i=480

Now yes, no dummy bars, no stupid comparisons. Straight forward and clean. That Hexus review seemed to be made by Jason Mick. Really.
August 19, 2008 3:05:03 PM

So Nehalem runs effectively 8 threads... but only gains between 15 to 30% over a same clock 4 thread C2Q on multi-thread software...

Hmmmm....
Related resources
Can't find your answer ? Ask !
August 19, 2008 3:18:43 PM

Yea, which all means zip to my benchmarks. Its like the ATOM, great for number crunching, but zilch regarding graphics usage. Im sorry, but Intel just doesnt get it. TY for the thread and quote, it helps, or really it doesnt. Oh well, theres always the next release, which will see gfx cards at least 50% faster by then, and thats if Intel hurries it up. And for those who think any comments negative towards AMD at this point wont help either. This isnt a Intel vs AMD thing. All this does is leave the door open for gfx cards to have plenty of headroom to do physx in games, because as they improve, theyll have to use that ability somewhere. And Intel said the gpu was dead. Man, they really dont get it
a b à CPUs
a b } Memory
August 19, 2008 3:19:11 PM

It doesn't have 8 physical cores, so why should it act like it does? What were you expecting?
August 19, 2008 3:23:44 PM

Higher clocks, better IPC
August 19, 2008 3:47:47 PM

cjl said:
It doesn't have 8 physical cores, so why should it act like it does? What were you expecting?


Clock per Clock ratio, Intel QuickPath Interconnect (IMC), updated architecture that can scale up to 80 cores , the return of Hyper-Threading, hype, and putting up with intel fanboys.

With all that we "were" expecting the best thing after sliced bread. And it seems to be called CUDA.
August 19, 2008 3:59:31 PM

cjl said:
It doesn't have 8 physical cores, so why should it act like it does? What were you expecting?


I would expect the machine running 8 instances to have a 15-30% advantage anyway...



I don't see much for IPC there.
August 19, 2008 4:14:18 PM

Im just let down is all. Servers are great, but as others have posted before me, who cares? As far as vid rendering goes, who kinows? Maybe itll show better than the quads out now, but theyre being challenged by CUDA. Multithreading? Only going to work in so many instances to begin with, then the software has to be adaptive to it. No, for me its all about IPC and higher clocks. Whats the max stock to be, at intro? Anyone know?
August 19, 2008 4:29:31 PM

JAYDEEJOHN said:
Im just let down is all. Servers are great, but as others have posted before me, who cares?


I care...


But I'm not convinced how useful hyperthreading is going to be for the kind of HPC work I do.



I guess the proof will be in the pudding... I'll just have to bide my time.



But hey, it may go some way to suggesting why Intel are pricing the thing so low. ;) 
August 19, 2008 4:35:53 PM

True that. Also, it should give AMD impetus to get off their arses and do a better desktop. AMD made their move with Barcelona, I guess were seeing Intel do theirs now. One of em are taking us DT users for granted. Just like the G280, it was good, nice increase in performance, costly, even now after price reductions. BUT, it wasnt what people were expecting in performance, it did "other" things, like physx and CUDA, and nVidia was more in tuned with Intel and Larrabee, they never saw ATI coming. Someone needs to pick up this gauntlet, and give us DT users some performance.
August 19, 2008 4:46:36 PM

Amiga500 said:
I care...


But I'm not convinced how useful hyperthreading is going to be for the kind of HPC work I do.



I guess the proof will be in the pudding... I'll just have to bide my time.



But hey, it may go some way to suggesting why Intel are pricing the thing so low. ;) 


For the mainstream market, it might come low priced. For your market, it will come expensive as hell. Ever looked at the prices of Xeon MP ? They are nice !!!

Hyper-Threading can seriously bump performance in some types of data. SQL/Oracle are a fine example. Small threads, but loads of them. Having worked with old Xeon Workstation (2x2.8 Mhz with HT) i can tell you for once, im happy with the return of a technology. HT didn't gave a massive bump in performance, but helped in case of hard locks and increased the stability (in performance and uptime). They were pretty fast machines for their time.

And improve data reads. Well, a picture is worth a thousand words.




August 19, 2008 4:59:34 PM

epsilon84 said:
http://www.hexus.net/content/item.php?item=15015&page=1

Check out the memory bandwith, insane! :ouch: 

Application performance ain't too shabby either, with Nehalem @ 2.93GHz generally beating the higher clocked QX9770 @ 3.2GHz, sometimes by huge margins. Gaming performance is poor, but can be attributed to improperly installed drivers apparently.



Funny that. I remember back in the days of X2 dominance that every reviewer would OC the heck out of Intel chips to get the scores higher, but with Phenom (now WITH SB750) no one OCs and shows how it compares to Intel. Am I just paranoid or are all reviewers full of it?

It's like they're not planning on showing what Phenom does at 3GHz+ until AMD actually releases that clock speed, but it doesn't help consumers to really see what they would be getting. All GPU tests are done with insanely expensive CPUs that no one can afford so none of us can know how the GPU will perform if WE upgrade.

Ever since Penryn, no sites are doing a full range of tests to show how CPU scales with GPU or how GPU scales with CPU. That's why I come to these sites. To see how much I need to spend to game at 1680x1050. If you're not showing that what are you doing?
August 19, 2008 5:02:46 PM

MarkG said:
So up to 35% faster clock-for-clock vs Intel's current generation in highly threaded applications and absolutely destroying AMD's current chips isn't good enough?

What exactly were you expecting?



I think we'd really need to show Phenom at 2.93GHz to really gauge. Especially since the 2.66Ghz i7 will be closest in price to 9950. What's the use in comparing chips that are separated by almost $300. The 2.93GHz was supposed to be $500+.
August 19, 2008 5:03:54 PM

BadTrip said:
how is Core2 crippled in 64-bit. Link?



Macro-fusion or whatever doesn't work in 64bit.
August 19, 2008 5:21:31 PM

BaronMatrix said:
Funny that. I remember back in the days of X2 dominance that every reviewer would OC the heck out of Intel chips to get the scores higher, but with Phenom (now WITH SB750) no one OCs and shows how it compares to Intel. Am I just paranoid or are all reviewers full of it?

It's like they're not planning on showing what Phenom does at 3GHz+ until AMD actually releases that clock speed, but it doesn't help consumers to really see what they would be getting. All GPU tests are done with insanely expensive CPUs that no one can afford so none of us can know how the GPU will perform if WE upgrade.

Ever since Penryn, no sites are doing a full range of tests to show how CPU scales with GPU or how GPU scales with CPU. That's why I come to these sites. To see how much I need to spend to game at 1680x1050. If you're not showing that what are you doing?


I'm with you on this one, BM. I *really* don't know what's the point of Anandtech reviewing the new SB750 while showing they can crank up the HT/NB speeds like hell, go past 3.0 GHz with the CPU and not posting a *SINGLE* damn benchmark. No, seriously.

Guess I'll try some underground forums. Or does anybody here have the new SB750 with a 3.0 GHz Phenom?
August 19, 2008 5:26:22 PM

dattimr said:
I'm with you on this one, BM. I *really* don't know what's the point of Anandtech reviewing the new SB750 while showing they can crank up the HT/NB speeds like hell, go past 3.0 GHz with the CPU and not posting a *SINGLE* damn benchmark. No, seriously.

Guess I'll try some underground forums. Or does anybody here have the new SB750 with a 3.0 GHz Phenom?



It's crazy. Some people on xtremesystems forum have some OC results. At 3.2GHz\2.4GHz L3, the Phenom is very close to the 9775 @ 3.2GHz. I'm not interested in whether it wins or loses, I just want as much info as I can get to determine for example if I can get away with a 9150e with a 4850 or if I should go to the 9550 and up.

Even AMDZone isn't showing OCd scores. WTH?
August 19, 2008 5:33:26 PM

AMD is still keeping us in the dark about the SB750. Or that or everybody is still in vacations. Except for me. Oh well......
August 19, 2008 5:55:21 PM

BaronMatrix said:
It's crazy. Some people on xtremesystems forum have some OC results. At 3.2GHz\2.4GHz L3, the Phenom is very close to the 9775 @ 3.2GHz. I'm not interested in whether it wins or loses, I just want as much info as I can get to determine for example if I can get away with a 9150e with a 4850 or if I should go to the 9550 and up.

Even AMDZone isn't showing OCd scores. WTH?


Well, I know that feeling of yours: it's disgusting. I'll be trying Google in a few minutes. But it's disappointing, nonetheless. Just finished reading an article at Nordichardware which says Nvidia will be updating its 780a to support ACC, but how the hell am I supposed to know if it's worth?

I'm torn between Intel's Q6600 and AMD's 9750, but I must say the 790GX platform looks amazing. THG is saying both 790FX and GX can support HyperTransport 3.1, so... It looks interesting.
August 19, 2008 6:21:17 PM

Bottom line is this. Nehelam is a server oriented cpu, and certainly not an upgrade for gamers, DT users. There is simply no benefit, so those that were intent on proclaiming Barcelona as a failed design, well have another look at Nehelam. Notice any similarities? So if you were in the "Barcelona sucks" category, then you must be squarely in the "Nehelam sucks" category as well.
August 19, 2008 6:32:59 PM

piesquared said:
Bottom line is this. Nehelam is a server oriented cpu, and certainly not an upgrade for gamers, DT users. There is simply no benefit, so those that were intent on proclaiming Barcelona as a failed design, well have another look at Nehelam. Notice any similarities? So if you were in the "Barcelona sucks" category, then you must be squarely in the "Nehelam sucks" category as well.



Ummm, that's blasphemy. How dare you compare i7 to K10. I just can't wait until people see how much i7 mobos will cost. X48 is inching up on $400. Striker is above it. X58 is supposed to require more layers. Imagine buying a $569 CPU with a $500 mobo and getting similar scores in high-res gaming as a person who buys a 9950 with a 790GX mobo.

Does that really HELP customers? Or how about forcing people to use G45 rather than an ATi 3200 level chipset. Sorry for the digression but I couldn't help it. Maybe AMDs lawyers read this site.
August 19, 2008 6:33:32 PM

piesquared said:
Bottom line is this. Nehelam is a server oriented cpu, and certainly not an upgrade for gamers, DT users. There is simply no benefit, so those that were intent on proclaiming Barcelona as a failed design, well have another look at Nehelam. Notice any similarities? So if you were in the "Barcelona sucks" category, then you must be squarely in the "Nehelam sucks" category as well.


This does not make sense, Nehelam just addresses cores weaknesses, How can the worlds fastest cpu suck?
August 19, 2008 6:34:55 PM

BaronMatrix said:
Ummm, that's blasphemy. How dare you compare i7 to K10. I just can't wait until people see how much i7 mobos will cost. X48 is inching up on $400. Striker is above it. X58 is supposed to require more layers. Imagine buying a $569 CPU with a $500 mobo and getting similar scores in high-res gaming as a person who buys a 9950 with a 790GX mobo.

Does that really HELP customers? Or how about forcing people to use G45 rather than an ATi 3200 level chipset. Sorry for the digression but I couldn't help it. Maybe AMDs lawyers read this site.



This just sounds like some people are hurting. Cry me a river.
August 19, 2008 6:42:15 PM

piesquared said:
Bottom line is this. Nehelam is a server oriented cpu, and certainly not an upgrade for gamers, DT users.


Desktop users will be happy with a five-year-old Athlon. Gamers who stick with their Core 2 Duos because 'no game uses more than two cores' will be whining because 64-bit games running 8 threads are much faster on Nehalem...

The improvement in IPC is a bit disappointing considering the amount they changed the core, but I'll certainly be buying one once the cost of DDR3 RAM drops to a sensible level.
August 19, 2008 6:45:15 PM

gallag said:
This does not make sense, Nehelam just addresses cores weaknesses, How can the worlds fastest cpu suck?


Remind me again of Intel's strategy. Is it Tick-Tock, or Tock-Tock? Because this definitley looks like a Tick-Tock-Tock. And judging from the very luke warm reception this is getting around the web, i'd have to say many agree.
And for the record, I never said it completely sucks. However it does look like only a minor improvement over Core. But heres the kicker, it's less of an improvement than Barcelona was over Rev F. So yeah, it looks like Intel is having trouble meeting their much talked about Tick-Tock strategy.
a c 126 à CPUs
a b } Memory
August 19, 2008 6:48:35 PM

You take it way too far man. The 3.2GHz QX9770 was there for comparison with the Core i7 to see where it stands.

WHat is it when AMD gets compared to Intels highest end everyone who prefers AMD gets up in arms and calls the price into question?

And to further continue, this review is not about the Phenom but the Core i7. Its also about seeing where Nehalem stands vs the highest end CPU you can get from BOTH parties.

Seriously calm the crap down and don't get defensive especially considering this may affect AMD as well.
August 19, 2008 6:51:05 PM

piesquared said:
Remind me again of Intel's strategy. Is it Tick-Tock, or Tock-Tock? Because this definitley looks like a Tick-Tock-Tock. And judging from the very luke warm reception this is getting around the web, i'd have to say many agree.
And for the record, I never said it completely sucks. However it does look like only a minor improvement over Core. But heres the kicker, it's less of an improvement than Barcelona was over Rev F. So yeah, it looks like Intel is having trouble meeting their much talked about Tick-Tock strategy.


Ahh, I see how I am getting confused, You are stating your opinion as fact. What is not to like about nehelam? about 10% ipc advantage over the current fastest and up to 40% in multi threaded apps. Nehalem will be a bigger push forward for Intel than deneb will be for AMD.

And what is this bs "And judging from the very luke warm reception this is getting around the web" the only people displaying this attitude are fanboy for the green team, If nehelam was a AMD cpu they would have to shake the jizz out of there pants.

August 19, 2008 6:58:39 PM

epsilon84 said:
http://www.hexus.net/content/item.php?item=15015&page=1

Check out the memory bandwith, insane! :ouch: 

Application performance ain't too shabby either, with Nehalem @ 2.93GHz generally beating the higher clocked QX9770 @ 3.2GHz, sometimes by huge margins. Gaming performance is poor, but can be attributed to improperly installed drivers apparently.


Fits right into the plan to get the core uArch to address server environments and HPC environments.

Word, Playa.
a c 126 à CPUs
a b } Memory
August 19, 2008 7:00:27 PM

BaronMatrix said:
Ummm, that's blasphemy. How dare you compare i7 to K10. I just can't wait until people see how much i7 mobos will cost. X48 is inching up on $400. Striker is above it. X58 is supposed to require more layers. Imagine buying a $569 CPU with a $500 mobo and getting similar scores in high-res gaming as a person who buys a 9950 with a 790GX mobo.

Does that really HELP customers? Or how about forcing people to use G45 rather than an ATi 3200 level chipset. Sorry for the digression but I couldn't help it. Maybe AMDs lawyers read this site.


You do realize that the mobos price is not decided by the chipset maker don't you? Its the mobo maker. Yes Intel chipsets have always been a bit more expensive but in comparison of quality (I am sure you have less experience since you wont buy Intel) compared to nVidia is worth the extra price. And I mean the nicley price mobos not the extreme haven't dropped yet mobos.

piesquared said:
Remind me again of Intel's strategy. Is it Tick-Tock, or Tock-Tock? Because this definitley looks like a Tick-Tock-Tock. And judging from the very luke warm reception this is getting around the web, i'd have to say many agree.
And for the record, I never said it completely sucks. However it does look like only a minor improvement over Core. But heres the kicker, it's less of an improvement than Barcelona was over Rev F. So yeah, it looks like Intel is having trouble meeting their much talked about Tick-Tock strategy.


Yet you have no server benchmarks to compare it to. That and the memeory bandwidth (BIG in server apps) that is being shown available from QPI is going to help Intel in the server market where AMD reigns supreme.

AMD better be worrying about what Nehalem will do in the server market. Not so much in the desktop market because their main money maker is in the server market.
August 19, 2008 7:07:49 PM

gallag said:
Ahh, I see how I am getting confused, You are stating your opinion as fact. What is not to like about nehelam? about 10% ipc advantage over the current fastest and up to 40% in multi threaded apps. Nehalem will be a bigger push forward for Intel than deneb will be for AMD.

?


And exactly, 10% ipc advantage. This was considered and outright failure with Barcelona.
Also, isn't 40% stretching it a bit? Unless you are seeing different benchmarks, the Hexus review shows 34% max. And that was in a benchmark in which it's predicessor failed. It's like having a foot race in which your competitor stumbles, while you run the same time, and claiming you run way faster than before.



Quote:
And what is this bs "And judging from the very luke warm reception this is getting around the web" the only people displaying this attitude are fanboy for the green team, If nehelam was a AMD cpu they would have to shake the jizz out of there pants.


Do you remember how the web was falling all over itself about Core 2? Well, look around. The forums i've seen with even a mention about it, have a smattering of resposnses at best. Those threads seem to migrate to the bottom of the page fairly quickly. So it's not the green team displaying attitude, it's more of a lack of interests all together, from what i'm observing.
a b à CPUs
a b } Memory
August 19, 2008 7:09:51 PM

I don't think it is underwhelming at all, except for people who had unrealistic expectations in the first place. It has a massive advantage in multithreaded and memory intensive applications, and a moderate advantage in single thread. That's exactly what they've been saying all along.
a c 126 à CPUs
a b } Memory
August 19, 2008 7:13:18 PM

^Um that was the difference between the Nehalem @ 2.93GHz and the Core 2 @ 3.2GHz. There is almost a 300MHz difference so the IPC is higher than 34%.

And it has been the "green" team that has been talking trash about a chip still not in its final form without fully finalized drivers.

They are also the ones getting mad about Phenom being in the same comparison as the 3.2GHz chip and this and that instead of just using it as a basis to compare with they moan about it being compared.

Hell the entire time Phenom was finally coming out people would always complain when they were comparing Phenoms performance to Core 2 Quad.
August 19, 2008 7:16:42 PM

piesquared said:
?


And exactly, 10% ipc advantage. This was considered and outright failure with Barcelona.
Also, isn't 40% stretching it a bit? Unless you are seeing different benchmarks, the Hexus review shows 34% max. And that was in a benchmark in which it's predecessor failed. It's like having a foot race in which your competitor stumbles, while you run the same time, and claiming you run way faster than before.



Quote:
And what is this bs "And judging from the very luke warm reception this is getting around the web" the only people displaying this attitude are fanboy for the green team, If nehelam was a AMD cpu they would have to shake the jizz out of there pants.


Do you remember how the web was falling all over itself about Core 2? Well, look around. The forums i've seen with even a mention about it, have a smattering of resposnses at best. Those threads seem to migrate to the bottom of the page fairly quickly. So it's not the green team displaying attitude, it's more of a lack of interests all together, from what i'm observing.


O.k then, Its only 34% faster & being better where predecessor failed is the whole point, Why do you see this as a bad thing, This is meeting expectation. This is what everyone has been expecting. Why does it hurt you so much?
August 19, 2008 7:32:49 PM

gallag said:
O.k then, Its only 34% faster & being better where predecessor failed is the whole point, Why do you see this as a bad thing, This is meeting expectation. This is what everyone has been expecting. Why does it hurt you so much?


au contraire amigo! It doesn't hurt at all.

Actually, from what i've been reading, the majority were expecting a massive improvement, like P4 to Core 2. So while i'm sure it most certainly met Intel's expectations, it doesn't appear to have met enthusiasts.
August 19, 2008 7:44:19 PM

piesquared said:
au contraire amigo! It doesn't hurt at all.

Actually, from what i've been reading, the majority were expecting a massive improvement, like P4 to Core 2. So while i'm sure it most certainly met Intel's expectations, it doesn't appear to have met enthusiasts.


O.K then, If this overestimating is not just in your head, Link to a few posts where people said it would be better than it is, For every one you find I will find ten that show that the general expectation has been achieved.

Every forum I have read I can not recall anyone guessing more than about 10% single threaded and 20-40% multi and server.
a b à CPUs
a b } Memory
August 19, 2008 7:44:47 PM

Plus we still have to see how it overclocks.
August 19, 2008 7:53:12 PM

Most People thought it wouldn't be as good as the phenom
since AMD was making this kind of cpu for awhile but everyone
was fooled.
without overclocking this cpu it beat the best Intel has to offer.
overclocking phenom can't beat the worst intel has to offer.
August 19, 2008 7:53:52 PM

gallag said:
O.K then, If this overestimating is not just in your head, Link to a few posts where people said it would be better than it is, For every one you find I will find ten that show that the general expectation has been achieved.

Every forum I have read I can not recall anyone guessing more than about 10% single threaded and 20-40% multi and server.


I'm not about to go on a link retrieving hunt. I'll get you started though. XS, [H], AT to name a few. In fact Johan has fairly good article up over at AT, where he explicitly states, this is not a gamers cpu. As a matter of fact it performs worse in that scenario, according to Hexus.


And when you do go on your link hunt, look through the threads and witness how many users have decided not to upgrade, in light of Nehelams performance. Like I said, this is a server cpu.
a b à CPUs
a b } Memory
August 19, 2008 7:57:43 PM

It performed worse because the drivers were wrong, not because it is inherently worse at gaming.
August 19, 2008 8:37:26 PM

cjl said:
It performed worse because the drivers were wrong, not because it is inherently worse at gaming.



That's possible cjl, as that's what Hexus has publicly stated. I just find it pretty hard to believe that Hexus somehow managed to sneakily get access to the Nehelam machine and run an hour or so worth of benchmarks, without Intel knowing about it. That seems pretty far fetched. You gotta think Hexus would be in pretty deep doo-doo if that were the case. Further, if Intel did grant Hexus access, why the heck would they have used preproduction hardware and software. When this thing is supposed to be all set to go, and release is imminent. At any rate, i'm willing to wait til official reviews are out to make any further claims.
August 19, 2008 8:38:36 PM

jimmysmitty said:
You do realize that the mobos price is not decided by the chipset maker don't you? Its the mobo maker. Yes Intel chipsets have always been a bit more expensive but in comparison of quality (I am sure you have less experience since you wont buy Intel) compared to nVidia is worth the extra price. And I mean the nicley price mobos not the extreme haven't dropped yet mobos.



Yet you have no server benchmarks to compare it to. That and the memeory bandwidth (BIG in server apps) that is being shown available from QPI is going to help Intel in the server market where AMD reigns supreme.

AMD better be worrying about what Nehalem will do in the server market. Not so much in the desktop market because their main money maker is in the server market.



true I haven't bought Intel but I can't find one post anywhere that says nVidia isn't as good. 790i is killing Intel. Th epoint was that if it costs x to make a mobo but 1.5x to make a Nehalem board, it will cost more. X48 is at close to $400. You can buy a 9850 AND a 790FX board for that price and with the same GPU have similar scores.

As far as Barcelona, AMD doesn't have to worry as a 2.3GHz K10 is beating a 2.93GHz Penryn (see Intel's benches). Shanghai @ 2.3 should perform like a K10 at 2.6 (10-15% higher IPC) which should perform like a 3.6GHz Penryn which doesn't exist. I guess though that a 3.2 i7 might get close to a 2.8 Shanghai but it will be MUCH BIGGER (did you see that pic compared to Penryn) and probably(not FUD) use more power. Plus it will require a full refit of a data center where Shanghai is a drop in replacement.


Again, I always say I hope they stay within 15% of each other so no one can be a fanboy.
a c 126 à CPUs
a b } Memory
August 19, 2008 9:11:47 PM

^You see the memory bandwidth in that review? Isn't memory bandwidth the biggest factor in server apps since most of them use a lot of the bandwidth and memory?

You are very confident that it will take a 3.2GHz i7 to take on a 2.8GHz Barcy yet there is yet to be a server i7/Xeon based on Nehalem put out by Intel to test soi how can you make a judgement?

Size wise and according to some earlier reviews it only used 10w more and the 3.2GHz is supposed to be a 130w TDP chip just like the 3Ghz Penryn was a 130wTDP (Yet it rarely hit that) of course we have to wait before we can make any judgements for a full review that goes over everything with drivers that are fully working and so on and so forth.
August 19, 2008 9:19:35 PM

jimmysmitty said:
^You see the memory bandwidth in that review? Isn't memory bandwidth the biggest factor in server apps since most of them use a lot of the bandwidth and memory?


The bandwidth is only necessary to keep your processing units fed.

If your actual number crunching units are sh!t, then bandwidth won't help much,


Moving bottlenecks.


On Penryn it is memory bandwidth. On Barcelona it is more likely processing speed itself.
August 19, 2008 9:25:07 PM

IDF nehelam news http://www.techreport.com/articles.x/15344

1:29PM: Nehalem demo of Lost Planet: Colonies. Yorkfield vs. Nehalem. Nehalem is over 50% faster thanks to eight cores, faster cores.

Now Cinebench with overclocked Nehalem and Yorkfiled. Over 30% faster on Nehalem. 45850 Cinebench R10 rendering score.
a b à CPUs
a b } Memory
August 19, 2008 9:29:56 PM

BaronMatrix said:
true I haven't bought Intel but I can't find one post anywhere that says nVidia isn't as good. 790i is killing Intel. Th epoint was that if it costs x to make a mobo but 1.5x to make a Nehalem board, it will cost more. X48 is at close to $400. You can buy a 9850 AND a 790FX board for that price and with the same GPU have similar scores.

As far as Barcelona, AMD doesn't have to worry as a 2.3GHz K10 is beating a 2.93GHz Penryn (see Intel's benches). Shanghai @ 2.3 should perform like a K10 at 2.6 (10-15% higher IPC) which should perform like a 3.6GHz Penryn which doesn't exist. I guess though that a 3.2 i7 might get close to a 2.8 Shanghai but it will be MUCH BIGGER (did you see that pic compared to Penryn) and probably(not FUD) use more power. Plus it will require a full refit of a data center where Shanghai is a drop in replacement.


Again, I always say I hope they stay within 15% of each other so no one can be a fanboy.

You keep calling X48 boards $400.

Lets' see how true that is:

Asus Rampage Formula = $299
DFI LP LT X48 = $250
Asus Rampage Extreme = $399
Gigabyte GA-X48 = $225
Intel reference X48 board = $250
Asus P5E deluxe X48 = $219
MSI X48C Platinum = $230


I'm only seeing one $400 board out of that sample from the top rated X48 boards on newegg. More importantly, the boards can easily be found in the $200 range (and a fairly decent selection too - not just a single cheap board). Likely, the X58 will be the same, with the top end boards coming in around the $400 mark, but easily attainable, and still very good boards being available in much lower price brackets.
August 19, 2008 9:49:12 PM

OK, I wont slam Nehalem because last time I came into the cpu section EVERYONE was saying it was waaaay faster than Penryn. Maybe its EVERYONES perception that has been turned down some? When Anand did his sneak preview, EVERYONE said the numbers were too small on the increases. They havnt changed much. It isnt Nehalem thats changed, its people thats changed. This is actually the first time Im hearing the lowered expectations. It was a different story when I came in and asked all my questions then. Go ahead and apologize for Nehalem all you want, but some of us were right from the get go, which I dont really care about so much as how my questions were perceived and treated. Stop with the love fest already. It is what it is. Good chip, not great. Has made great strides where AMD was winning, and left DT somewhat out of it, so, me being a DT user, and a gamer, Im still disappointed, since now we know how itll perform.
August 19, 2008 10:37:28 PM

jaydee, does your Penryn chip run games too slow for you or something? I dont understand the whine fest your putting on...
August 19, 2008 10:50:15 PM

OK, so now its a whine fest? Have you read my posts? Will Nehalem keep up with gpus? Before I answer you, you answer me. Im not talking now, and Im not talking Penryns. Im talking about Nehalem. Im also talking about the only path Intel is selling is multithread, which hasnt done alot for any of us. And some people here are programmers, and devs. They get crapped on, why? Is it because, as usual Intel wants it their way? What about a truly faster cpu? One that will keep up with gpus in speed? If this is whining, then youve bought into something, or believe this is the only way. Its a way for Intel to make more money. If they started designing chips that went faster instead of wider, which would you buy? Is that whining also? To ask that question? Answer those questions for yourself.
August 19, 2008 11:02:23 PM

What? Keep up with GPUs? What the heck does that have to do with anything? That is pretty ridiculous. Tell us, jaydeejohn, what CPU has kept up with GPUs right now? That is a very lame argument. What does the speed of a GPU have to do with anything? Did you ask the same question when Phenom was being launched?

How about this question - Does Nehalem keep up or surpass the previous CPU generation it will probably replace? Yes or no?

Also, I did a search for your so-called Intel fanboys and their "higher" expections, and I couldn't find anyone claiming any huge performance increase with Nehalem vs. Penryn. Here's a search of "Nehalem" in these forums. Can you find the high fanboy expectations, that you claim EVERYONE said?
http://www.tomshardware.com/forum/forum1.php?recherches=1&cat=28&orderSearch=0&config=tomshardwareus.inc&pseud=&search=Nehalem&titre=1&jour=18&mois=5&annee=2008&resSearch=200&daterange=2&subcat=174&searchtype=1&trash=0&trash_post=0&moderation=0

Keep up with GPUs? Wow. Talk about reaching for excuses to make something look bad.

Quote:
In fact Johan has fairly good article up over at AT, where he explicitly states, this is not a gamers cpu.

So, what does it mean when the gaming benchmarks surpassed those of the Phenom 9950 BE? That it is a better gamer CPU than the Phenom it was compared to?
!