Multi-GPU SLI/CF Scaling: Lynnfield's Blemish

When running in single-GPU mode, the on-die PCIe controller maintains a full x16 connection to your graphics card:


Hooray.

In multi-GPU mode, the 16 lanes have to be split in two:

To support this the motherboard maker needs to put down ~$3 worth of PCIe switches:

Now SLI and Crossfire can work, although the motherboard maker also needs to pay NVIDIA a few dollars to legally make SLI work.

The question is do you give up any performance when going with Lynnfield's 2 x8 implementation vs. Bloomfield/X58's 2 x16 PCIe configuration? In short, at the high end, yes.

I looked at scaling in two games that scaled the best with multiple GPUs: Crysis Warhead and FarCry 2. I ran all settings at their max, resolution at 2560 x 1600 but with no AA.

I included two multi-GPU configurations. A pair of GeForce GTX 275s from EVGA for NVIDIA:


A coupla GPUs and a few cores can go a long way

And to really stress things, I looked at two Radeon HD 4870 X2s from Sapphire. Note that each card has two GPUs so this is actually a 4-GPU configuration, enough to really stress a PCIe x8 interface.

First, the dual-GPU results from NVIDIA.

NVIDIA GeForce GTX 275 Crysis Warhead (ambush) Crysis Warhead (avalanche) Crysis Warhead (frost) FarCry 2 Playback Demo Action
Intel Core i7 975 (X58) - 1GPU 20.8 fps 23.0 fps 21.4 fps 41.0 fps
Intel Core i7 870 (P55) 1GPU 20.8 fps 22.9 fps 21.5 fps 40.5 fps
Intel Core i7 975 (X58) - 2GPUs 38.4 fps 42.3 fps 38.0 fps 73.2 fps
Intel Core i7 870 (P55) 2GPUs 38.0 fps 41.9 fps 37.4 fps 65.9 fps

 

The important data is in the next table. What you're looking at here is the % speedup from one to two GPUs on X58 vs. P55. In theory, X58 should have higher percentages because each GPU gets 16 PCIe lanes while Lynnfield only provides 8 per GPU.

GTX 275 -> GTX 275 SLI Scaling Crysis Warhead (ambush) Crysis Warhead (avalanche) Crysis Warhead (frost) FarCry 2 Playback Demo Action
Intel Core i7 975 (X58) 84.6% 83.9% 77.6% 78.5%
Intel Core i7 870 (P55) 82.7% 83.0% 74.0% 62.7%

 

For the most part, the X58 platform was only a couple of percent better in scaling. That changes with the Far Cry 2 results where X58 manages to get 78% scaling while P55 only delivers 62%. It's clearly not the most common case, but it can happen. If you're going to be building a high-end dual-GPU setup, X58 is probably worth it.

Next, the quad-GPU results from AMD:

AMD Radeon HD 4870 X2 Crysis Warhead (ambush) Crysis Warhead (avalanche) Crysis Warhead (frost) FarCry 2 Playback Demo Action
Intel Core i7 975 (X58) - 2GPUs 25.8 fps 31.3 fps 27.0 fps 70.9 fps
Intel Core i7 870 (P55) 2GPUs 24.4 fps 31.1 fps 26.6 fps 71.4 fps
Intel Core i7 975 (X58) - 4GPUs 27.0 fps 57.4 fps 47.9 fps 117.9 fps
Intel Core i7 870 (P55) 4GPUs 24.2 fps 50.0 fps 36.5 fps 116 fps

 

Again, what we really care about is the scaling. Note how single GPU performance is identical between Bloomfield/Lynnfield, but multi-GPU performance is noticeably lower on Lynnfield. This isn't going to be good:

4870 X2 -> 4870 X2 CF Scaling Crysis Warhead (ambush) Crysis Warhead (avalanche) Crysis Warhead (frost) FarCry 2 Playback Demo Action
Intel Core i7 975 (X58) 4.7% 83.4% 77.4% 66.3%
Intel Core i7 870 (P55) -1.0% 60.8% 37.2% 62.5%

 

Ouch. Maybe Lynnfield is human after all. Almost across the board the quad-GPU results significantly favor X58. It makes sense given how data hungry these GPUs are. Again, the conclusion here is that for a high end multi-GPU setup you'll want to go with X58/Bloomfield.

A Quick Look at GPU Limited Gaming

With all of our CPU reviews we try to strike a balance between CPU and GPU limited game tests in order to show which CPU is truly faster at running game code. In fact all of our CPU tests are designed to figure out which CPUs are best at a number of tasks.

However, the vast majority of games today will be limited by whatever graphics card you have in your system. The performance differences we talked about a earlier will all but disappear in these scenarios. Allow me to present data from Crysis Warhead running at 2560 x 1600 with maximum quality settings:

NVIDIA GeForce GTX 275 Crysis Warhead (ambush) Crysis Warhead (avalanche) Crysis Warhead (frost)
Intel Core i7 975 20.8 fps 23.0 fps 21.4 fps
Intel Core i7 870 20.8 fps 22.9 fps 21.5 fps
AMD Phenom II X4 965 BE 20.9 fps 23.0 fps 21.5 fps

 

They're all the same. This shouldn't come as a surprise to anyone, it's always been the case. Any CPU near the high end, when faced with the same GPU bottleneck, will perform the same in game.

Now that doesn't mean you should ignore performance data and buy a slower CPU. You always want to purchase the best performing CPU you can at any given pricepoint. It'll ensure that regardless of the CPU/GPU balance in applications and games that you're always left with the best performance possible.

The Test

Motherboard: Intel DP55KG (Intel P55)
Intel DX58SO (Intel X58)
Intel DX48BT2 (Intel X48)
Gigabyte GA-MA790FXT-UD5P (790FX)
Chipset: Intel X48
Intel X58
Intel P55
AMD 790FX
Chipset Drivers: Intel 9.1.1.1015 (Intel)
AMD Catalyst 9.8
Hard Disk: Intel X25-M SSD (80GB)
Memory: Qimonda DDR3-1066 4 x 1GB (7-7-7-20)
Corsair DDR3-1333 4 x 1GB (7-7-7-20)
Patriot Viper DDR3-1333 2 x 2GB (7-7-7-20)
Video Card: eVGA GeForce GTX 280
Video Drivers: NVIDIA ForceWare 190.62 (Win764)
NVIDIA ForceWare 180.43 (Vista64)
NVIDIA ForceWare 178.24 (Vista32)
Desktop Resolution: 1920 x 1200
OS: Windows Vista Ultimate 32-bit (for SYSMark)
Windows Vista Ultimate 64-bit
Windows 7 64-bit

Turbo mode is enabled for the P55 and X58 platforms.

The Best Gaming CPU? SYSMark 2007 Performance
Comments Locked

343 Comments

View All Comments

  • TA152H - Wednesday, September 9, 2009 - link

    Your assumption would be wrong, I use computers for the normal stuff people do, and compiling.

    Go back to picking your nose, lowlife.
  • Skiprudder - Tuesday, September 8, 2009 - link

    I'm starting to feel like some folks have Bloomfields and now they're trying to justify spending the money they did. At no point did Anand say Bloomfields were 'bad', he's just pointing out that due to current price/performance ratios the new chips are fantastic for the vast amount of folks here. They really area big step up for a lot of people, and frankly a lot of us feel we have better things to spend out money on than $300 X58 mobos. We can now get in on that sort of performance (or better) for a reasonable price, and how isn't this great?
  • TA152H - Tuesday, September 8, 2009 - link

    Well, you might be right, but within this context I'm going to assume you are talking about me, in which case you would be wrong.

    I don't have anything new, and I'm actually not going to get anything for a few months. Even if I did, I wouldn't let it cloud my perspective. Normally people who make these judgments are looking through their own personality flaws. I have many, but this is one area I do not.

    Maybe I am annoyed at the processor because I was expecting more. I was expecting it to be a really nice processor, and it turned out disappointing. I get annoyed with stupid things sometimes, like the 4 clock cycle L1 cache irks me big time with the Nehalem. With the P55, the platform is just weak and I don't really like it when sites do everything they can to obfuscate the compromises in it.

    I'll say this, though, about it, the power use is REALLY nice. I'm really impressed with that. Everything else though, just leaves me cold.

    There are some questions that need to be asked, as well. Why is the P55 so damn expensive? It's the same cost as the P45, but without most of the logic.

    Why is the 870 so expensive? Is there really any point to this processor at this cost? Maybe some, but not really for a broad segment of the buying population. I'd much rather have an LGA1366 if I were going into that type of expense, instead of the brain-damaged derivative.

    LGA1366 motherboards are around $200 now, at least many are, so $300 is kind of an outdated number. For this, you get better i/o, better memory flexibility (you can use two or three dimms; you can't use three with a P55), better performance, etc...

    Now, one thing Anand brought up, and I didn't respond to, was something about not needing more bandwidth unless you were tapping out all four processors. Needless to say, this is obviously false, and I wonder why he'd repeat it. A little thought would tell us that even with two cores running, you could gain advantages by having better bandwidth. The reason is simple. Let's say Core A and Core B are both plugging away, and Core A needs a memory read. OK, so it takes the bus, and starts the long process of getting a cache line read. Now, let's say a clock cycle, or two, or twenty, later, Core B needs memory outside of cache. So now it needs the memory bus. Hmmmm, but Core A has the bus. So, Core B has to wait. If you have three memory channels, you transfer data faster, satisfying the cache line faster, and freeing the bus sooner. So, now Core B can get it, and start working sooner. So, you could see better performance with lower processor use requirements than they indicate.

    Now, the only possibility this premise is wrong is if the core only grabs 64-bit memory bank at a time, which seems very unlikely to me. In this event, there would still be one memory channel open to the second processor, thus no penalty would occur until three processors needed memory. I don't think this is what they did, as it would lower memory performance on the Bloomfield unless three processors were being used, and also, based on the shared L3 cache, it seems the entire memory bus is always used. Still, it is possible. Maybe Anand can answer this.
  • mesiah - Wednesday, September 9, 2009 - link

    Dude, you are so full of shit they can smell it 3 states away. Don't you dare try to rationalize by saying "I was expecting more." Since lynnfield news first broke you have done nothing but trash the part, Call it brain damaged, and skew posted facts to fit your warped agenda. Then, when the retail part is finally unveiled, instead of eating crow (something you suggest Anand do) you act like every crazy half baked conspiracy theorist that ever existed and start ranting about biased reporting, facts being intentionally hidden, and flawed logic.

    While I will admit that this isn't the be all end all lynnfield review, this is day one of the launch. I believe more information was displayed in this review than can be reasonably expected for the initial review. In depth overclocking comparisons and the like will certainly come later, but that isn't to say overclocking was ignored.

    You read the article and then go off on a tirade about biased reporting and the skewing of facts to meet an agenda when it is you that have been skewing facts all along, and you continue to do so. You make overclocking comparisons between lynnfield and bloomfield calling them similar, then ask why anyone would give up "all the other advantages" of the bloomfield if you are overclocking, but you fail to mention the big advantage for most, which is price comparison. Then later you try to rationalize price by quoting the lowest posted 920 sale price next to the 870 msrp. The parts hit the shelves today, give them a month for prices to normalize. You aren't going to get a smokin deal on the first one to roll off the production line.

    Now, go ahead and call me a fool like you do everyone else that is smart enough to see you for what you are. Attempting to verbally abuse another person in order to make yourself feel smarter is a pretty common tactic for feeble minded people who are themselves afraid of looking stupid. Its the equivalent to bringing a gun to a fist fight just in case you start to get your ass beat. So, how about you stop being a pussy, eat a little crow, and admit that all of your talk about how terrible this part was going to be was wrong. Or better yet, just stop posting here because your pissy "I'm better than you" attitude does nothing but bring the site down.

    You can reply and call me all the dirty names you want, you wont get a response. I only feed the trolls once a week.
  • TA152H - Wednesday, September 9, 2009 - link

    Actually, you're an idiot, and you're changing my words to create an argument.

    I never said it was terrible, I said the pre-release performance was terrible, and I expected it to be better for the released version. It was, but I still don't think so highly of the processor, and I wouldn't consider it. I'd rather get the i7 920.

    If you're going to argue, at least have the decency/intelligence to not misrepresent someone's position.

    I didn't really read the rest of your crap. I could only skim through your sub-literate drivel. I doubt you would have said anything useful in it.
  • Skiprudder - Thursday, September 10, 2009 - link

    Can you please stop referring to people as idiots and morons? I'm not sure where you acquired your rhetorical skills, but calling folks names is the last thing one should do if you're actually trying to convince people of your viewpoint. Call someone here a moron, and it makes you look like just a troll. Calm down, stay rational, and people will be much more likely to hear what you have to say (and if they aren't, why should you care anyway?) There's no excuse for rudeness.
  • Skiprudder - Tuesday, September 8, 2009 - link

    I agree, the 870 isn't priced at all sensibly (but I would argue the top Bloomfields are pretty darn unreasonable too). Prices are rather high right now, and it will be interesting to see what they do over the next few weeks as supplies and demands start to balance out.
  • chrnochime - Tuesday, September 8, 2009 - link

    The sad thing is, right now(at least), in order to get i7 CPU and decent MB to OC with, a P55 set up would run:

    ~300 for the i7 860
    plus
    ~200 for an UD4P or an Asus P55 Pro.

    Whereas for a x58 setup,

    I can get an i 7 920 for 200 pre-tax, and should spend about 230-250 for a decent motherboard.

    So for me, it's actually cheaper to go with the x58 setup, even though P55 MB are supposed to be cheaper...
  • TA152H - Tuesday, September 8, 2009 - link

    Yes, and you'd end up with better performance too. And better i/o, and better flexibility.

    Once you factor in overclocking, the P55 isn't really much of an option unless you go to it's really low end, where you simply can't build an x58. Then it's competing with the Core 2, and has a chance.
  • ClownPuncher - Tuesday, September 8, 2009 - link

    Good article, thanks for the clarifications too.

Log in

Don't have an account? Sign up now