Phenom II/LGA775 review with 5000 series Cards?

[sincreator]

Just wondering why there are no reviews covering the performance of the Ati 5800 series cards with a Phenom II/LGA775 used for the CPU. I know that the I7 should be a better suited CPU for testing these cards, but there are alot of people using the Phenom II/LGA775 chips, and it would be pretty interesting to see how much of a bottle neck they actually would present while gaming. Single and crossfire results.

Right now I am running an AM3 x3 710 at 3.45ghz, and I'm trying to decide if I should get the 5850/5870. I also have the ASUS M3A32-MVP(quad crossfire capable MOBO) 790fx. So eventually I will be looking to crossfire 2 of the cards in question. Basically I don't want to waste money if there will be a huge CPU bottleneck. If there will be a big bottleneck, I may wait to get the second card, and just get a 6-core PH II after christmas. Any one got a link to such a review? I dont care about synthetic benchmarks, I want to see game benchmarks. Perferably comparing results with I7 vs. Phenom II vs. LGA775 at the same clockspeeds, with and without crossfire with the 5850, and the 5870.

PS: I play on my 32" LCD TV at 1920x1080. If that helps at all. lol

EDIT: I also posted this question in the video card section, but later I realized that it probally would of been better to post it here. So sorry for the double posting. I also added the LGA775 Cpu's to the discussion since alot of the people here are using them. (My brother has an e-8500 at 4.28ghz, and he wanted to know as well)

 

[bige420]

A single 5870/5850 probably wouldn't cause a huge bottleneck, but CF two of them together would probably cause a noticeable bottleneck.

 

[sincreator]

That's why it would be nice to see a review with a Phenom II/LGA775 using crossfire with these cards. I can't find any reviews anywhere with this information, which is quite surprising. Come on Tom's, this is a chance to one up all the other sites out there. The i7 probally only accounts for 20%(or less) of your readers, and it would be nice to know for future upgrade decisions. If needed or not. If the i7 will only gain 5-10% over a core2quad/core2duo/Phenom/Phenom II, maybe people could hold off for a bit longer to get a bigger performance increase from a future CPU.

 

[dirtmountain]

I agree with you 100% sincreator. It seems all the reviewers have gone with the i7 9xx/LGA1366 platform for almost all testing of new graphic cards due to the ease of testing for them. It pretty much leaves those of us with different system platforms (AM3/LGA775) out in the cold.

 

[jsh1284]

Well although this is far from being a legitimate review, but I don't think I'm experiencing any bottlenecking issues with my 955 PII and my 5870. Speeds - amazing. Reliable - perfectly and awesomeness - FTW.

 

[Wisecracker]

I think the Asus M3A32-MVP is a great motherboard but it most likely is holding back your OC (because of the sb600) and will eventually limit your expansion (the mobo is approaching 2 years old and I would imagine BIOS updates will become more and more limited).

I also think that you Phenom 710 may partially impact an HD5xxx Crossfire but at 3.4GHz probably not as much as you think --- there will still be a big benefit from the second card (and as always, the card drivers may make or break the experience).

 

[sincreator]

I think the Asus M3A32-MVP is a great motherboard but it most likely is holding back your OC (because of the sb600) and will eventually limit your expansion (the mobo is approaching 2 years old and I would imagine BIOS updates will become more and more limited).

I also think that you Phenom 710 may partially impact an HD5xxx Crossfire but at 3.4GHz probably not as much as you think --- there will still be a big benefit from the second card (and as always, the card drivers may make or break the experience).

Oh for sure the sb600 is holding me back. With a 700 series south bridge I should be able to get 3.6ghz 100% stable(maybe higher). At one point I had 3.6ghz working pretty stable, but it would crash after 3/4 hours of constant playing, so I dropped back to 3.45ghz for 24/7 use. My ram is also DDR-2 800mhz default, but I have that OC'ed to 1066mhz as well. TBH Asus have been supporting this board better than I expected(maybe they sold alot of them), so I figure that the new 6 core phenoms will be supported as well. It is still 16x16 pci-e 2.0 for 2 slot crossfire. It's 16x8x8 for 3 way, and 8x8x8x8 for 4 way. I will never go above 2 cards though, but I may grab a cheap Nvidia card for Physx down the road. That's if it gets 100% supported while using ATI for the main display.

 

[jj463rd]

I think that you would do well with 2 Radeon 5850 cards in crossfire.2 Radeon 5870 cards would be too much.I also have a M3A32-MVP motherboard.I am thinking about getting another Phenom II X4 955 BE CPU as an upgrade when they come down in price though.

 

[keithlm]

Oh for sure the sb600 is holding me back.

Actually that is not always true... the PhII chips don't always gain a lot from the ACC. (Other than perhaps allowing you to unlock the 4th core.)

With my Foxconn A79A-S motherboard my PhII 940 could do 3.8-3.9Ghz. With my current Gigabyte MA790GP-UD4H it can only do 3.7-3.8Ghz. I recently tested tried my old Gigabyte MA790FX-DS5 because I wanted to test the difference between x16/x16 and x8/x8 and found that it could do 3.9-4.0Ghz with the newest bios. (Those speeds are all with Vista 64. I could probably get 4.2-4.3Ghz with XP.)


So in reality it appears that I got slightly better overclocking on the SB600 than on the SB750.

Sadly though the Gigabyte DS5 board is not very stable. (It's a known "lemon". They never officially supported the 9850 or 9950 chips even though the board can handle the power... it is the general consensus that Gigabyte had a hardware issue they couldn't fix with a bios update.)

SIDE NOTE: The x16/x16 results were slightly faster than x8/x8, but not enough to really matter. Besides I'm probably going to remove my third 4850 soon. It's only 512Mb and my 4850x2 card has 1Gb for each of it's GPU. But using all three GPU drops them all to 512Mb. And some newer games actually do use the additional memory.

 

[sincreator]

With the amount of voltage tweaks that I needed for the 3.45ghz stable OC, I would say it was the SB600 holding me back on this board. Allthough my particular chip also needed 1.475 v-core to stay stable at 3.45ghz, and I have read it should only need about 1.4v for that OC. At the same time my HT-Link and NB freq. needed to be below 2150 as well. I've read that the boards with the sb700/750 chipset will run up to 2600 usually. Now maybe that was just for the quads, I'm not 100% sure about that.

About what you were saying about the amount of ram on your cards. I always thought that the second, third, and forth card's memory didn't do anything in crossfire. I thought it was just the first card's ram that ever got utilized. Correct me if I'm wrong.(Wouldn't be the first time. lol )

 

[keithlm]

With the amount of voltage tweaks that I needed for the 3.45ghz stable OC, I would say it was the SB600 holding me back on this board. Allthough my particular chip also needed 1.475 v-core to stay stable at 3.45ghz, and I have read it should only need about 1.4v for that OC. At the same time my HT-Link and NB freq. needed to be below 2150 as well. I've read that the boards with the sb700/750 chipset will run up to 2600 usually. Now maybe that was just for the quads, I'm not 100% sure about that.

About what you were saying about the amount of ram on your cards. I always thought that the second, third, and forth card's memory didn't do anything in crossfire. I thought it was just the first card's ram that ever got utilized. Correct me if I'm wrong.(Wouldn't be the first time. lol )

Have you tweaked the CPU-VDDA voltage on your board while overclocking the cores and/or the NB? If not you will want to check into that; you will need to increase that from stock.

For purposes of the CORE and overclocking: If you use a SB750 motherboard and leave the ACC disabled in the BIOS the board is acting exactly the same as if you had a SB600. (You do get the added SATA etc.)

With my PhII 940 I have not seen any benefit from using ACC; I get the same overclock and voltage requirements with it enabled or disabled. In fact the only difference I have seen on my system is less stability with it enabled. It is possible I haven't found the magic settings, but I can get to 3.5Ghz on stock voltages with ACC disabled and 3.6-3.7Ghz with a small voltage increase. (Then it takes a lot to get to 3.8Ghz or higher.)

You have a good motherboard. It sounds like you got a questionable PhII chip. I suspect if you put in a brand new MB with a SB750 you would be disappointed in how that chip works on the new motherboard also. But then maybe not... it is often said that ACC works best with problem chips and does almost nothing with a "golden" chip.


About Crossfire: see: http://www.pcper.com/article.php?aid=523 for the quote: (Of course he is talking about mixing 256Mb and 512Mb but the same applies.)

Driver will always go to lowest common denominator whether it will be memory speed or frame buffer. So in essence the 512MB card will be seen as a 256MB card. I unfortunately don’t have scaling numbers handy for 256 and 512 MB cards mix-ups (I don’t actually think we ever would have benchmarked that scenario).

 

[JAYDEEJOHN]

PCGamesHW did a review with a i7 920 at 3.5 with turbo off using CFd 5870s, and yes, in some games, even the 920 was limiting the fps, so yes, you AMD rig will too, but all games showed decent fps, and minimal fps were fine, where you really need both card and cpu to come thru.
If youre sporting a decent 3+Ghz cpu in CF with these cards, theyre so powerful it really wont matter, yet, tho in future games, the DX11 option should help with cpu restrictions

 

[sincreator]

@KeithLM

I wasn't sure how to have this as a quote, so I just copy pasted this from here: http://www.tomshardware.com/forum/245454-33-crossfire-faqs

"Do SLI or CrossFire double the memory ?
This is one the questions that many people make mistake understanding it .
The answer is NO , SLI or CrossFire doesnt double the memory , You have a 512MB card , adding another card wont make your memory 1GB, it still will be 512MB
Another example :
A game needs 512MB card to run at maximum settings , and you have a 256MB card , and you think if you add another 256MB card , your memory will be 512MB and thats not true."

I know that the GPU's will run at the slower cards specs, but the memory is not doubled when running crossfire. So why would the second cards ram even matter if it's not being used at all. It should be that if the first card has 1gb of ram and the second has 512, the second cards memory shouldn't even be a factor.

 

[sincreator]

PCGamesHW did a review with a i7 920 at 3.5 with turbo off using CFd 5870s, and yes, in some games, even the 920 was limiting the fps, so yes, you AMD rig will too, but all games showed decent fps, and minimal fps were fine, where you really need both card and cpu to come thru.
If youre sporting a decent 3+Ghz cpu in CF with these cards, theyre so powerful it really wont matter, yet, tho in future games, the DX11 option should help with cpu restrictions

Any idea on if there will be a review done in the future using AM3/LGA775 with the 5800 series cards? Like I was saying before, I know that right now the I7 is probally the best CPU for testing a GPU's full potential, but it would be very helpfull to the people that are not running I7 CPU's decide on what card to get depending on their current hardware. After the problems that Nvidia GPUs had performance wise with the I7 compared to the Phenom II, I think it may be worth a try at the very least. http://www.tomshardware.com/reviews/geforce-gtx-280,2156.html (Getting benchmarks right: Editors corner)

 

[keithlm]

So why would the second cards ram even matter if it's not being used at all. It should be that if the first card has 1gb of ram and the second has 512, the second cards memory shouldn't even be a factor.

You have to understand that memory IS used on each and every card in a Crossfire system; it is not just disabled on all cards other than the first one.

The system "mirrors" the memory on each and every card. In order to mirror the same memory on all cards it is limited to the size of the smallest card. (EDIT: I don't think that's exactly how it works... but it is basically what the end result is.)

(Actually it would be very nice if only the size of the memory on the first card was relevant.)

 

[JAYDEEJOHN]

Toms has been the only consistent review site doing both cpus, as they did with the CF/SLI recently with 2 4890s on a AMD rig.
Unfortunately, it just doesnt happen the often.
If I come across one, I'll try n post it here, so anyone can check it out, and I want to make sure its a decent review as well, as some sites dont do credible work, as evidenced by not having this done, as per your question, it all comes down to money, and some cut corners, some are biased, and some just dont review cards well

 

[sincreator]

Toms has been the only consistent review site doing both cpus, as they did with the CF/SLI recently with 2 4890s on a AMD rig.
Unfortunately, it just doesnt happen the often.
If I come across one, I'll try n post it here, so anyone can check it out, and I want to make sure its a decent review as well, as some sites dont do credible work, as evidenced by not having this done, as per your question, it all comes down to money, and some cut corners, some are biased, and some just dont review cards well

That was why I posted this question here. I knew that Tom's and Anandtech were the only sites that had done this recently. They drew the same conclusions about Nvidia cards on i7 as opposed to the phenom II as well. Which is why I thought it would be really interesting, even if the i7 destroys the Phenom II with the 5000 series cards. It would just be good to know.

I've been looking for a review covering AM3/LGA775 effect on the 5800 series performance since the 5800 series was released, with no luck at all. It would be absolutly fantastic if you can find one and post it here. So many people could find it very usefull. The thing that seems strangest to me is that this is an AMD/ATI video card, and no one has thought to review it with an AMD cpu. Very weird if you ask me, since it is a part of their overall platform.

 

[keithlm]

In reviews with older multiple card setups we see the Phenom II doing just fine on a bunch of benchmarks and then one or two specific benchmarks it shows problems so the reviewer immediately concludes that the Phenom II is bottlenecking.

Later (in different reviews) when something similar happens to an i7 the reviewer immediately concludes that there must be driver issues.


So yes it will be definitely be interesting to see the results of how the PhII does with the new 5000 series. These new tests will either support the previous bottlenecking myth or it will put that myth to rest once and for all.

 

[sincreator]

Who know's what will happen with the Phenom/LGA775 CPU's? Since it isn't being tested anywhere, how do we know that the Phenoms/LGA775 will run into the same limitations that the I7's have with the 5800 series GPU's? I mean AMD made the Phenoms and the 5000 series cards, and maybe they were able to optimize them to work better together. We won't know for sure until someone tests it. It almost seems that this is not being done on purpose for some reason.
Basically, its showing all these cpus are the slowdowns, as only the oceed i7 actually brings more out of the cards, with the cards not using any AA.
Im thinking, when we look at a x2 card from this gen, the cpu will be the bottleneck half the time in most games, unless oceed, and maybe still even if oceed

 

[JAYDEEJOHN]

Heres a small quick review from a decent site. Its just using the 5870 only, and its main purpose is to show scaling with cpus. While the only oceed cpu was the i7 here, Im sure the i5 wouldve done just as well, if not better.
The difference between the cpus is somewhat as we see with other tests, as Intels IPC (Instructions Per Clock) is higher than AMDs, and it shows in these tests
http://www.pcgameshardware.de/aid,697048/Radeon-HD-5870-im-Test-CPU-Skalierung-mit-6-CPUs-in-WoW-Risen-Anno-1404-FC2-und-Crysis-Warhead/Grafikkarte/Test/
However, the AMD solution is beating the 775 here, as also already known. Things may change once W7 comes out, and uses more of the newer setup available on the OS

 

[sincreator]

Thanks, this is what I've been looking for. I figured that this triple core cpu would hold back a 5870. I guess I'll just get one 5850 for now and a second one down the road, after my next CPU upgrade. The CPU upgrade is probally going to be a 6-core phenom next year, since it will be a drop in replacement.

PS: Why was half of the moderators post in my last message? Forum problem?

 

[JAYDEEJOHN]

Whooops, I was tired, hit the wrong button, sorry. Late nighter