Sign in with
Sign up | Sign in
Your question
Closed

GeForce And Radeon On Intel's P67: PCIe Scaling Explored

Last response: in Reviews comments
Share
a b å Intel
April 11, 2011 4:26:52 AM

These tests were done with a single card, on X16/X8/X4 slots. Fine.
But... Who would use anything other than a X16 slot if they had one?
The only real use for a X8 slot would be for sli/crossfire where the addition of a second card should result in an Increase of performance, not a decrease.
April 11, 2011 4:30:48 AM

it would be nice if you included the GTX 570 in the x8/x8 and x16/x4 test. the 570 flies over the 6950 on the single card config but i was curious how it would do since i think the AMD's scale better than nvidia's. would the 570 still lead because of its advantage? or would AMD even it out because of its scaling (if any) "advantage"
Related resources
April 11, 2011 4:32:36 AM

^They already did an article on that man
April 11, 2011 4:49:51 AM

Can you guy's do an article on how performance is affected if you SLI/Xfire using PCI-E 16x slots running @ 4x?

3 way would be preferable because if performance is still adequately faster I'll consider it.
April 11, 2011 6:05:13 AM

Good comments. Can we please see 8x/8x and 16x/4x since that comparison is relevant? I get the impression that somehow SLI/crossfire reduces the performance hit of x4 lanes but I'd like to see numbers.
April 11, 2011 6:25:03 AM

joytech22Can you guy's do an article on how performance is affected if you SLI/Xfire using PCI-E 16x slots running @ 4x?
I think you missed a page then!
http://www.tomshardware.com/reviews/pci-express-scaling...
The numbers were there all along!

carlhenryit would be nice if you included the GTX 570 in the x8/x8 and x16/x4 test.
Well, you should probably read the linked page too then. There's no point in artifically creating a configuration (by taping lanes or whatever) that doesn't exist in real life, is there?

"While Nvidia prevents SLI from functioning on PCH-hosted lanes, x16/x4 configurations are completely possible in CrossFire. But should they be? We tested our motherboard in both x8/x8 and x16/x4 configurations to find out."
April 11, 2011 6:41:25 AM

Yeah...my bad.
a c 134 å Intel
April 11, 2011 8:27:16 AM

Looking forward to see the third article!
April 11, 2011 9:15:42 AM

Im guessing that 8x lanes are mostly enought. I do belive that it would depend on how fast an actual gpu is, as well as how much ram it has, and how big is its bandwidth.
I means, its logical, but mayb not true :D . Would be nice to see this test on a GTX560 Ti, since it has a lot of headroom for OC, then compare oced version vs non oced. Also this might be interesting in GPUs that have diffrent versions with more and less RAM.
Just my 2 cents :D .
April 11, 2011 9:57:37 AM

cats_pawIm guessing that 8x lanes are mostly enought. I do belive that it would depend on how fast an actual gpu is, as well as how much ram it has, and how big is its bandwidth.I means, its logical, but mayb not true . Would be nice to see this test on a GTX560 Ti, since it has a lot of headroom for OC, then compare oced version vs non oced. Also this might be interesting in GPUs that have diffrent versions with more and less RAM.Just my 2 cents .
It's...different for sure. Just about anything that can "bottleneck" graphics performance will have a greater effect on AMD GPUs than on Nvidia GPUs. CPU too slow? DRAM? PCIe? It's going to hurt Radeons more than GeForces.

Of course, Tom's Hardware has all the cool kit that helps editors overcome those bottlenecks.
April 11, 2011 10:39:38 AM

Finally the x16/x4 question answered!! It got me thinking tho, since ATI allows "mismatched" cards do it matter which slot the faster card is in......
April 11, 2011 11:09:43 AM

Great article!
April 11, 2011 11:15:38 AM

feeddagoatFinally the x16/x4 question answered!! It got me thinking tho, since ATI allows "mismatched" cards do it matter which slot the faster card is in......



that's an interesting question, like a 6950 and a 6970, would it be faster with a 6970 in the first pcie slot or the second.
April 11, 2011 11:33:24 AM

I'd love to know what AMD cards are doing that requires so much more bandwidth... faster memory speeds, perhaps?

It might be worth conducting at least two way Crossfire/SLi tests more often in future with new card reviews or, at least, follow-up reviews, though I can imagine it'll delay such articles a little. Perhaps using AMD cards on P67 is limiting their scaling, even if it's already very good (with the 6000 series)? Would using two mid-range cards have the same effect?
April 11, 2011 11:50:57 AM

AHHHH! Good info. Can't wait for the next part!!!
April 11, 2011 12:04:19 PM

One thing though... No BFBC2 performance charts?

This is a mighty popular game. Does EA make reviewers pay them to mention their game in reviews?
April 11, 2011 12:39:48 PM

it depends on your chipset if you want awsome scaling get a motherboard with a better chipset
April 11, 2011 1:39:40 PM

One nitpick are some of the conclusions, like x16/x4 don't work well for AMD, or NVIDIA. It could be correct, but you haven't proven anything in this article. Although, one could now make that case for the LGA 1155/1156 platforms, and this certainly goes a long way in showing those deficiencies, you'd still have to prove this isn't platform related, or extenuated. Your assumption is being four lanes wide is the problem, but we'd have to prove that it's not related to where the lanes are attached to before we could fully prove it's the width.

For example, would this be the case on LGA 1366, where all the PCIe lanes are on the chipset, and not some going through the chipset and some being part of the CPU package? In real world situations, it's not so important because you'd have less cause to use an x4 in LGA 1366, but for academic purposes, and a better understanding of what's really going on, it could be interesting.
April 11, 2011 2:26:48 PM

So, i currently have a 5770 on w/a i5-750, obviously w/1156 chipset, would it be a good idea to get another 5770. Reason i ask is i have 2 pci express slots but second one is at 4x lanes. I have a asus p7p55d le motherboard. Do i have to buy same card or could it be a different brand at different speeds. Or better to get the 6950 2gb i want to buy.
April 11, 2011 2:30:31 PM

silverblueI'd love to know what AMD cards are doing that requires so much more bandwidth... faster memory speeds, perhaps?


I too am wondering why this is. My non-educated guess would be that the AMD card has more memory, requiring more bandwidth. I wonder if there would have been such a sizable hit on the 1GB version of the 6950 when in the x8 or x4 slot.

And thank you Tom's for answering the x16/x4 Xfire question that a lot of us had. However, it would have been nice to see the single x16 card's numbers on the same chart to illustrate the moderate gains an additional card in the x4 slot gives us. Thanks again for another great article.
April 11, 2011 2:34:32 PM

Well, I would have thought the drop would be more on these cards compared to the last time we saw this with a 5870. Makes me wonder where I'll go with a future card upgrade...

I would like to see 16x/16x vs 8x/8x done more thoroughly though. Say an X58 vs P67 (NF200) vs P67 (regular).
April 11, 2011 3:05:39 PM

Do really got to say, this kind of test should have been done on both a Intel and AMD based platform. Especially the x8/x8 and x16/x4 test. For that matter, would also have been good to see not just higher end cards used, but also cards like the GF 460 and HD 5770, so that GPU performance can be factored into how much bandwith is needed per card per pci-e slot.

That way it could be shown whether or not the lower end cards also take just as much of a performance hit at x8 and x4 as the higher end cards do, or if they do not get held back nearly as band. Reason I say this is that prior test have been done on prior graphics cards before using x4 and also x16/x4 dual card modes, and while showing a performance impact, it was no where near as bad as what you have shown. One would definitely assume that the newer generation higher end cards are indeed going to demand more from their pci-slots, as they can do more. But is this ultimately the same case (as of now anyway) with the more mainstream cards from the Geforce 450, 460, 550 line, along with the Hd 5770, 5830, 6790, and 6850 line?

Ultimately many factors are going to come to play. One is the games themselves. F1 2010 and Just Cause 2 for example took major performance hits in your test, even in Crossfire x16/x4. The other titles seemed to take less of a impact however, depending on AA settings, ect, with fps loss of maybe 4-7 fps depending on said title, showing that performance loss will definitely vary per title, with some titles still benefiting quite well from Crossfire regardless of the second card running at x4.


Another factor, as stated, is what platform is being used. You did not include any AMD based platform in this test, nor any more commonly used mainstream cards. Due to this the scope of your test was far too controlled by limiting factors, and only provided quite limited results that honestly can not speak for every x16 through x4 configuration, let alone every x8/x8 and x16/x4 dual card configuration, even with a margin of error provided. In other words, this is hardly what I would consider the end all of articles on pci-e scaling explored. Sorry Tom, on this one, no dice.
April 11, 2011 3:24:55 PM

I thought Tom's was better than this. The PCIe speeds only matter for data transfers to/from the card. Once a game/level is loaded there is much much less data going back and forth (usually), unless.....

You choose a graphics card with alot less memory causing the game to fill the graphics card memory and rely on system memory for the overflow of textures and geometry. Then you should see much bigger variations between the 4x/8x/16x slots.

My opinion is the tests you are doing are flawed as all the games textures and geometry fit within the graphics card memory and the only real traffic thru PCIe is the updates to geometry/shaders/etc...nothing to big.

Maybe benchmark the difference in load times too.
April 11, 2011 3:27:03 PM

let this be a lesson for all those "know it alls", as well as the newbies looking for advice here. Don't believe everything you read on these forums. So many people pulling "knowledge" out of thin air, and while it sounds good, it leads people astray.
I know people who changed their minds on which motherboard to get based on the erroneous information provided by "veterans" here.
Thanks Tom's for speaking out on this.
April 11, 2011 3:33:55 PM

geofeltThese tests were done with a single card, on X16/X8/X4 slots. Fine. But... Who would use anything other than a X16 slot if they had one?The only real use for a X8 slot would be for sli/crossfire where the addition of a second card should result in an Increase of performance, not a decrease.


Well, me for one. Space is tight on a uATX board, and my H67 mobo has 1 16x, one 16 @ 4x, and one 1x (as well as a useless PCI). With Asus Essence STX soundcard and a Killer 2100 network card, I had to shuffle some stuff around. I'm using a much more modest GTX 460 Cyclone, and my own testing showed much less of a hit in all the games I've tried. There was almost no drop in framerates, though most games I play have much lower requirements (like Fallout: NV).

If you can, you should use a 16x slot. If you need the flexibility of placement, a 8x or 4x can work. Just test it out first.

Good job, Toms.
a b å Intel
April 11, 2011 3:48:16 PM

Good point.
April 11, 2011 3:51:13 PM

what about a test of video cards and systems for Adobe CS5 with 64bit, all you doing is all the time videocards for games, I don't care about games but don't want to buy the wrong card or spend to much $$$ for a semi pro


Thanks
a b å Intel
April 11, 2011 4:03:49 PM

Exactly what I was expecting ..... but really want to see the follow up article on SLI / CF. As performance increases enthusiasts deal with the law of diminishing returns ....as we get closer to the highest level, we have to pay more and more for a 1% performance increase.

On a typical $2k box, for example, moving from sat The Asus P8P67 Deluxe to the WSW Revolution adds $20 or 1% to your system cost. An investment of 1% that returns a 2% in performance is, to my mind a good investment for peeps using nVidia cards. That 2:1 rate of return jumps to 4:1 for ATI users.

Of course, if your instead leaning toward a P8P67 Pro, that's a $70 investment, and let's assume a $1400 box this time. That's an additional 5% for a 2 - 4 % improvement which brings us back to the law of diminishing returns. Each buyer has to make the decision on whether the law is something they can live with.
April 11, 2011 4:18:19 PM

klausjpwhat about a test of video cards and systems for Adobe CS5 with 64bit, all you doing is all the time videocards for games,


Ummm, because maybe 1% of the readers here use CS5?? LAst I heard CS5 dev's were not considered enthusiasts... just buy a card and move along...

I thought the article was pretty good. I know each reader wants their gpu or mb tested, but that actually doesnt tell anyone anything. It is very hard to reduce the variables to answer the question - which in this case was scaling between a couple of GPU's and X16, X8 and X4 PCIe slots.

There may be a couple of follow on articles about Intel vs AMD scaling, or additional gpus but this was one was well done.
Thanks guys
April 11, 2011 4:18:35 PM

I have an AM2+ board with x16 and x4 slots. Now, I know I can't use Crossfire on this board thanks to the presence of Hybrid Crossfire, however if a 4x slot reduces performance as much as this, I'm rather glad I didn't get that second 4830.

AMD FX setups feature 38 2.0 lanes, though they'll probably be CPU bottlenecked in general, so it might be interesting to try various combinations on an overclocked X58 setup i.e. 16x/16x, 16x/8x/8x/8x etc. etc., though I'm sure it'd take a good while to run through!
April 11, 2011 4:19:16 PM

is x16/x4 in crossfire better than just one x16 though? or will x16/x4 actually decrease performance from just 1 x16?
a b å Intel
April 11, 2011 4:36:28 PM

Interesting, although I'm another who wasn't too surprised by the results. Three questions I'd like to see explored are:
1. Why are the assorted performance hits greater for AMD cards? This one might be a question for AMD engineers, and I doubt they'd be all that candid, unless the reason is a worthwhile sacrifice to achieve superior performance in other ways / on other tests.
2. Are the results obtained with AMD boards similar, and if not, why not?
3. On an AMD-chipset board using a nVidia GPU, if a second one is added for PhysX (SLI being unavailable), what are the effects on PhysX performance of using a X4 slot?
April 11, 2011 4:42:11 PM

Crossfire is superior to one 16x card, regardless of setup, or at least as far as I can see from a quick look at the benchmark results.
April 11, 2011 4:45:04 PM

Refering to the crossfire of x8/x8 versus x16/x4:
After noticing Tom's didn't seem recommend the x16/x4 configuration, I decided to review their findings. While I agree the F1 game didn't scale very well for AMD compared to all the other games, let's assume this is driver related, so if we take the average numbers from this one game out of the average, the hit for going from x8/x8 to x16/x4 would change as follows:

1680x1050: 6.3% instead of 10.3%
1920x1080: 5.7% instead of 9.6%
2540x1600: 5.7% instead of 10.9%

While I agree this is still a considerable margin, people that already have this type of board (x16/x4) considering crossfire, it seems to me it would be far more economical to purchase a second video card, rather than upgrading their motherboard AND buying a second video card. So, they would save $200 on a mobo purchase, and lose out by about 4% performance (10%-6%).
To put in perspective, say I have 40fps with one card, 75fps with two cards at x8/x8, or 70fps with two cards x16/x4, I would be happier with the $200 in my pocket than have those 5fps.
Does this make sense or am I missing something?



April 11, 2011 5:59:55 PM

Great article, finaly proof for a younger brother who swears you need 16 Lanes or else it's crap. I always advised people that X8/X8 is OK, but X16/X4 would leave one card very un-used. I always recommend to people who plan to crossfire steer clear of the 1156 boards that only do x16/x4.
April 11, 2011 6:40:52 PM

Noodletoe said:
Refering to the crossfire of x8/x8 versus x16/x4:
After noticing Tom's didn't seem recommend the x16/x4 configuration, I decided to review their findings. While I agree the F1 game didn't scale very well for AMD compared to all the other games, let's assume this is driver related, so if we take the average numbers from this one game out of the average, the hit for going from x8/x8 to x16/x4 would change as follows:

1680x1050: 6.3% instead of 10.3%
1920x1080: 5.7% instead of 9.6%
2540x1600: 5.7% instead of 10.9%

While I agree this is still a considerable margin, people that already have this type of board (x16/x4) considering crossfire, it seems to me it would be far more economical to purchase a second video card, rather than upgrading their motherboard AND buying a second video card. So, they would save $200 on a mobo purchase, and lose out by about 4% performance (10%-6%).
To put in perspective, say I have 40fps with one card, 75fps with two cards at x8/x8, or 70fps with two cards x16/x4, I would be happier with the $200 in my pocket than have those 5fps.
Does this make sense or am I missing something?


It makes sense, and I agree with the second part of your arguement. I have a x16/x4 board and if I need more graphics power I will add another 6950 into the mix before I get an x8/x8 mobo....if only for the reason that it will take me about two months to purchase both. With a 2nd GPU, I see the results as soon as I install the card.

As for the first part of your post, I'm not a fan of throwing out results because they don't look good. Driver related or not, the end user takes a serious frame rate hit. While I don't play F1, I will assume these results are comparable to different games out there that Tom's doesn't include in their testing and should be included to give a better (more objective) result.
April 11, 2011 7:29:37 PM

TA152HOne nitpick are some of the conclusions, like x16/x4 don't work well for AMD, or NVIDIA. It could be correct, but you haven't proven anything in this article. Although, one could now make that case for the LGA 1155/1156 platforms, and this certainly goes a long way in showing those deficiencies, you'd still have to prove this isn't platform related, or extenuated. Your assumption is being four lanes wide is the problem, but we'd have to prove that it's not related to where the lanes are attached to before we could fully prove it's the width.For example, would this be the case on LGA 1366, where all the PCIe lanes are on the chipset, and not some going through the chipset and some being part of the CPU package? In real world situations, it's not so important because you'd have less cause to use an x4 in LGA 1366, but for academic purposes, and a better understanding of what's really going on, it could be interesting.
Au contraire! http://www.tomshardware.com/reviews/p55-pci-express-sca...
It's already been looked into several times for other chipsets, including the X58.
silverblueI have an AM2+ board with x16 and x4 slots. Now, I know I can't use Crossfire on this board thanks to the presence of Hybrid Crossfire, however if a 4x slot reduces performance as much as this, I'm rather glad I didn't get that second 4830.AMD FX setups feature 38 2.0 lanes, though they'll probably be CPU bottlenecked in general, so it might be interesting to try various combinations on an overclocked X58 setup i.e. 16x/16x, 16x/8x/8x/8x etc. etc., though I'm sure it'd take a good while to run through!

I'm fairly sure you can pick the cards you want to use for CrossFire.
rajohns08is x16/x4 in crossfire better than just one x16 though? or will x16/x4 actually decrease performance from just 1 x16?

You get your performance increase, in a smaller dose.

April 11, 2011 8:11:40 PM

Simply good, good article but I why don't you used a PLX enabled motherboard for that comparison, IMHO you lost one to make ten.
April 11, 2011 9:25:52 PM

CrashmanI'm fairly sure you can pick the cards you want to use for CrossFire.


Yes, but most of the attention will (rightfully?) be on using higher end cards... and in any case, my board (GA-MA78G-DS3H rev. 1.x) definitely cannot take Crossfire in any shape or form besides the Hybrid variety. :(  I'll leave that for my next setup, I think... :) 

I'm mainly interested, personally, in low- to mid-range cards for both AMD and NVIDIA and how they fare in this sort of comparison.
April 11, 2011 9:51:48 PM

silverblueYes, but most of the attention will (rightfully?) be on using higher end cards... and in any case, my board (GA-MA78G-DS3H rev. 1.x) definitely cannot take Crossfire in any shape or form besides the Hybrid variety. I'll leave that for my next setup, I think... I'm mainly interested, personally, in low- to mid-range cards for both AMD and NVIDIA and how they fare in this sort of comparison.
Are you really certain of that? I've never ran across ANY limitation on CrossFire, and the board does have two slots that can hold graphics cards. You could simply disable the onboard graphics and run two cards, CrossFire should work for those two cards.
April 11, 2011 10:05:14 PM

Question, does anyone have the link to the 16x/16x vs 16x/8x vs 8x/8x done on TH awhile back? If so, please send and thanks!
April 11, 2011 10:16:49 PM

Great review as always
Though I have a question.I read some articles about the comparison between x16x4 and dual x8 on PCI-E 1.0 boards(with 2 of either PCI-E 2.0 and 1.0 cards installed) and the difference between them was much more than what it is on PCI-E 2.0 boards (with 2 PCI-E 2.0 cards installed)
SO why is the performance drop is lower on a PCI-E 1.0 board compared to PCI-E 2.0 boards?
April 11, 2011 11:12:26 PM

MaziarGreat review as always Though I have a question.I read some articles about the comparison between x16x4 and dual x8 on PCI-E 1.0 boards(with 2 of either PCI-E 2.0 and 1.0 cards installed) and the difference between them was much more than what it is on PCI-E 2.0 boards (with 2 PCI-E 2.0 cards installed)SO why is the performance drop is lower on a PCI-E 1.0 board compared to PCI-E 2.0 boards?
Different hardware and different benchmarks. Some games need more bandwidth. ATI cards need more bandwidth. Newer ATI cards need even more bandwidth when tied to high-bandwidth games. And newer/higher-frequency processors shift the bottleneck away from CPU, towards PCIe.
April 12, 2011 2:16:12 AM

I have done a mod where I have 1 spare PCI-E x1 slot on my mobo. I cut the back from to insert a GeForce 430 x16 card. My primary slot uses a GeForce 570 SuperClocked to near 580 spec. I have set the x1 slot to be used for Physx ONLY and this frees up the primary card to get over 10 fps better average at 1900x1200 in Crysis2.

beware... I tried using a blade at first... then quickly decided to dremel with a thick metal piece protecting the mobo. This is a dangerous thing to tinker with!
April 12, 2011 4:02:56 AM

In the next article, you guys should address performance hit taken by a multi-GFX card, multi-monitor setup when switching from x16/x16 to x8/x8, as it is quite significant. This is what has kept me from upgrading my long-in-the-tooth Q9550 to a 2600k. I'm waiting for LGA2011 or Bulldozer.
April 12, 2011 6:41:27 AM

CrashmanAre you really certain of that? I've never ran across ANY limitation on CrossFire, and the board does have two slots that can hold graphics cards. You could simply disable the onboard graphics and run two cards, CrossFire should work for those two cards.


Well, I did a search a while back and Google seemed to suggest that it couldn't be done, however I did find a nice review this morning at Bit-Tech which actually features 4870s in Crossfire on that board in comparison with the 790G chipset. Seems to work quite well really, even if the review is somewhat skewed due to the use of a triple-core Phenom against a quad-core, and the number of games they test is rather limited.

I think the question now is whether it'd be a worthwhile performance boost, especially considering I only have a PII X3 710, or if I should really wait and build a new setup in the not-too-distant future. Regardless, you made me look into it further... thanks! Very happy to be wrong on this point. :) 
April 12, 2011 6:44:17 AM

Well its good to know that 16x/4x is not that good now.

Will you test a AMD system with low cost M/B like 780g or some 880g. Most of them offer CrossFire with 16x/4x PCIe, because i think is not fair there are only intel p67 there.

How about the Mid-Range Cards like 5770 or GTS 450 in CrossFire and SLi.

maybe this setup will do some good

AM3 x6 CPU
ASUS M4A88TD-V
2 x 5770
!