GeForce GTX 480 PCI-Express Scaling

[W1zzard]

NVIDIA's latest beast offers unprecedented performance to the enthusiast gamer. But it also requires massive amounts of data being fed to it to perform. That's why it is commonly expected that high-end graphics cards are crippled if the PCI-Express interface does not offer the maximum speed of x16 2.0. We test four different PCI-E configurations and their performance to shine light on this question.

Show full review

 

[Steevo]

I see.

 

[jasper1605]

thanks

 

[SystemViper]

i like the insight into the 8x vs the 16x pcie slots, again anothter sweet review/test

 

[Apocolypse007]

So little difference between 8x and 16x. I guess thats a good thing though. If you were limited to running a card in 8x for SLI, one might be concerned about a performance hit due to limited bandwidth.

 

[theorw]

Great review!The kind u wont find anywhere else in the net!
So 2% difference 16x>8x doesnt justify the price of x48 and u we should be just fine with 2x 8x of p45 right?
Thought so too since i xfired on both and saw no difference!!

 

[TAViX]

Excellent. So using 8x it's not such a big deal afterall... Good for SLI/Crossfire thingy

 

[HalfAHertz]

Is it possible to test Sli and CF at x16 vs x8? I think the hit will be much more than 2% due to the increased bandwidth.

 

[nt300]

Is it possible to test Sli and CF at x16 vs x8? I think the hit will be much more than 2% due to the increased bandwidth.

There real is no hit with 16x and 8x whether you in crossfire/sli or not. Check benchmarks, 780GX vs. 790FX with what I seen the FX was slitly faster but still bad in price/performance when compare to GX.

 

[HillBeast]

Interesting that. I was expecting performance to halve when in less lanes but the drop was pretty minimal. Cheers for the insight.

 

[theorw]

Is it possible to test Sli and CF at x16 vs x8? I think the hit will be much more than 2% due to the increased bandwidth.

There no decrease u can notice while gaming...Benching maybe but only when using say 2x 5970 etc...
I have used both X48 and P45 chips on my 4850s and didnt notice a thing different in performance!

 

[SKD007]

WHy are you using 9.12 for ATI ??
Here is a fud version

http://www.fudzilla.com/content/view/18242/1/

and they used a 10.3a

but can some one compair with similar config system ??

 

[Mac Daddy]

First off what an insightful review I had no idea that the difference between x16 and x8 was almost negligible. Great job and I read through the one on the 5870 as well

Is it possible to test Sli and CF at x16 vs x8? I think the hit will be much more than 2% due to the increased bandwidth.

I had a similar question but mine is directed more towards the bandwidths themselves in SLI or Crossfire. As far as I have read in either configuration it is ideal to match the core and memory clocks on all the cards used.

My question is running SLI or Crossfire in a x16 x8 configuration or x16 x8 x8 configuration going to affect performance and stability?

 

[DonInKansas]

WHy are you using 9.12 for ATI ??

This was already covered in another thread. Leave it be.

 

[btarunr]

My question is running SLI or Crossfire in a x16 x8 configuration or x16 x8 x8 configuration going to affect performance and stability?

No issues with stability, driver irons that out. Go ahead with a x16 x8 x8 build.

 

[Mac Daddy]

No issues with stability, driver irons that out. Go ahead with a x16 x8 x8 build.

Thanks for the quick reply. I have been shopping for a mobo for an i7 test system and have been holding off because everything seems to be x16 x8 x8. I was looking at this mobo actually.

http://www.gigabyte.com.tw/Products...rd&ProductID=2957&ProductName=GA-EX58-EXTREME

 

[dumo]

Its gonna be a lot of HD5870 for sale soon

 

[Solaris17]

This was already covered in another thread. Leave it be.

 

[crazyeyesreaper]

now this was very informative i love the PCIe scaling reviews on the new GPUs tells me exactly what i need to know for crossfire or sli on lower end boards and it seems x4 is plenty of bandwidth so a x4 x16 board can still sli and get good results from the looks.

thanks again W1zz

 

[Nick89]

Its gonna be a lot of HD5870 for sale soon

Only idiots will sell there 5870's for a 12% increase from the GTX480. you can get a 12% increase from your 5870 by overclocking it.

 

[[I.R.A]_FBi]

dugg

 

[boulard83]

Only idiots will sell there 5870's for a 12% increase from the GTX480. you can get a 12% increase from your 5870 by overclocking it.

+1 ... sry +950 !

I was able to run 950/1300 on the stock cooling. Whit water in it its a breeze ...

This review is interesting but i think someone received some money for this review

 

[HillBeast]

now this was very informative i love the PCIe scaling reviews on the new GPUs tells me exactly what i need to know for crossfire or sli on lower end boards and it seems x4 is plenty of bandwidth so a x4 x16 board can still sli and get good results from the looks.

thanks again W1zz

Correct me if I'm wrong, but the PCI-E bus has the most stress in loading when it has to send the textures to the card, and then after that it is mainly instructions. I'm probably wrong, but I'd say this is why there would be such a minimal drop.

 

[lemonadesoda]

Correct. You are not wrong. Minimum framerates are directly impacted by this, not just complexity of the scene being rendered.

This review is very interesting, but what it misses in the discussion is the interaction of PCI-E bandwidth, memory bandwidth, and chipset. The reason is that x16 2.0 is getting very close to memory bandwidths, give or take all the overhead and data marshalling.

W1z is getting some difference in performance on his highly OC rig where memory bandwidth >> PCI-E bandwidth, so that it is the PCI lane that is causing a potential bottleneck.

BUT for a regular DDR2 or DDR3 non-OC extreme edition, the difference would be much smaller still. So while it is nice to see OC figures to the max, it may also misrepresent the true behaviour for the regular user.

Furthermore, if you have SLI crossfire system, the results WILL NOT scale in the same way. Why? Because the bandwidth needed to fill 2x PCI-x16 2.0 lanes simultaneously is getting pretty close to memory bandwidth, esp. with all the data-marshalling going on in the chipset.

I would like to encourage w1z to do a follow-up to his otherwise most excellent review:

1./ I would really like to see the results at stock.
2./ Do a test with SLI/crossfire. The whole x8 x16 discussion is moot without putting into real context. The results will not scale in the same way as the single card set up... so therefore the results of the single card setup are irrelevant to the discussion about SLI/Crossfire performance under x4, x8 or x16
3./ By all means, do a sensitivity analysis on how OC'ing improves the results, or indeed worsens them from a relative perspective
4./ But also see how the results change by increasing/decreasing memory bandwidth by dropping a memory channel or two
5./ As 1 but do the test on a similarly clocked dual xeon system with memory channel bandwidth to the max, and see how much the results pull away compared with 1.

I think the results 5 vs. 1 vs. the original would be a very interesting comparative and help enthusiasts see the value of dual xeons vs single CPU vs x4 x8 x16 vs. SLI/Crossfire.

 

[lemonadesoda]

The poor man has to sleep too! Never mind about follow up analysis. The points are valid though.

To flog an old horse, if a GTX480 works so well on a PCI-E x4, then what about AGP... it would have been fine too, possibly better performance than the x4 even, due to lower latencies

There's a great review of SLI GTX480 over on anantech. Wow, the new GT100 architecture scales well. But OMG look at those SLI power consumptions, over 800W system power with a simple SLI setup! eek!