5870 Bottleneck Investigation (CPU and/or Memory Bandwidth)

[jaredpace]

Unigine Heaven Benchmark - 1680x1050 - 4xAA, 16xAF - max w/tesselation - Cat 9.10

650/900 = 23.9 fps
650/950 = 24.4 fps
650/1000 = 24.7 fps
650/1050 = 24.9 fps
650/1100 = 25.1 fps
650/1150 = 25.2 fps
650/1200 = 25.5 fps
650/1250 = 25.7 fps
650/1300 = 25.8 fps

725/900 = 26.0 fps
725/950 = 26.2 fps
725/1000 = 26.7 fps
725/1050 = 27.0 fps
725/1100 = 27.3 fps
725/1150 = 27.4 fps
725/1200 = 27.8 fps
725/1250 = 28.0 fps
725/1300 = 28.2 fps

800/900 = 27.7 fps
800/950 = 28.2 fps
800/1000 = 28.6 fps
800/1050 = 28.9 fps
800/1100 = 29.3 fps
800/1150 = 29.6 fps
800/1200 = 29.8 fps
800/1250 = 30.1 fps
800/1300 = 30.4 fps

875/900 = 29.4 fps
875/950 = 29.9 fps
875/1000 = 30.2 fps
875/1050 = 30.7 fps
875/1100 = 31.1 fps
875/1150 = 31.4 fps
875/1200 = 31.8 fps
875/1250 = 32.1 fps
875/1300 = 32.4 fps

950/900 = 30.7 fps
950/950 = 31.1 fps
950/1000 = 31.7 fps
950/1050 = 32.1 fps
950/1100 = 32.6 fps
950/1150 = 33.3 fps
950/1200 = 33.6 fps
950/1250 = 33.8 fps
950/1300 = 34.4 fps

mem vs core

900/650 = 23.9 fps
900/725 = 26.0 fps
900/800 = 27.7 fps
900/875 = 29.4 fps
900/950 = 30.7 fps

950/650 = 24.4 fps
950/725 = 26.2 fps
950/800 = 28.2 fps
950/875 = 29.9 fps
950/950 = 31.1 fps

1000/650 = 24.7 fps
1000/725 = 26.7 fps
1000/800 = 28.6 fps
1000/875 = 30.2 fps
1000/950 = 31.7 fps

1050/650 = 24.9 fps
1050/725 = 27.0 fps
1050/800 = 28.9 fps
1050/875 = 30.7 fps
1050/950 = 32.1 fps

1100/650 = 25.1 fps
1100/725 = 27.3 fps
1100/800 = 29.3 fps
1100/875 = 31.1 fps
1100/950 = 32.6 fps

1150/650 = 25.2 fps
1150/725 = 27.4 fps
1150/800 = 29.6 fps
1150/875 = 31.4 fps
1150/950 = 33.3 fps

1200/650 = 25.5 fps
1200/725 = 27.8 fps
1200/800 = 29.8 fps
1200/875 = 31.8 fps
1200/950 = 33.6 fps

1250/650 = 25.7 fps
1250/725 = 28.0 fps
1250/800 = 30.1 fps
1250/875 = 32.1 fps
1250/950 = 33.8 fps

1300/650 = 25.8 fps
1300/725 = 28.2 fps
1300/800 = 30.4 fps
1300/875 = 32.4 fps
1300/950 = 34.4 fps

http://forums.techpowerup.com/showth...106814&page=15

[jaredpace]

I think this says compute power @ core of 650mhz gets a good increase as memory bandwidth increases. Also, quite interestingly, as compute power of the core increases towards 1000mhz, the performance increase associated with the memory bandwidth increase becomes greater itself. (degree of performance increases as well!).

[Smartidiot89]

Strange idea:

Anyone tried changing the PCI-E mhz anything and see if it affects the performance?

[Blkout]

Strange idea:

Anyone tried changing the PCI-E mhz anything and see if it affects the performance?

I got a small performance boost from raising the PCI-E from 100 to 115MHz when I used my 285GTX, but I get no gain when I change it with my 5850's.

[***Deimos***]

No, I think the memory timings are static, they do not change with memory adjustment. So if they are at, for example, a total cycle delay of 10, that would mean:

10 cycles @ 1200MHz = 8.3ns delay
10 cycles @ 1300MHz = 7.7ns delay

Because there is no memory bottleneck on the 5870 cards, when you increase the memory speed, all you are really doing is decreasing the nanosecond delay on memory operations, which is why a 10% increase in memory speed is only getting a 1.5% increase in performance.

Here is good table of specification comparison of GDDR3, GDDR4 and GDDR5.
http://theovalich.wordpress.com/2008...45-diferences/

1. GDDR5 has high latency and so Fermi/Cypress have lots of cache/buffers. Unless special ultra high resolution with very large texturing benchmark is used, the mem system isn't CONSTANTLY used.

2. Although higher mem clock does marginally improve performance, maybe because offset by increasing errors.

3. Remember the i7/Phenom "un-core" clock. You only see significant mem scaling when you also increase the GPU clock... perhaps because this is increasing the mem controller speed and cache. ie doesn't help speeding up getting data to the chip, when the chip is slower and takes time carrying data where needed.

4. Driver optimizations have far greater impact than ALL OTHER FACTORS COMBINED. A well tuned shader compiler which makes good use of registers and cache can make minimal use of memory bandwidth - of course the behaviour would be different for other vendors/cards.

5. Issue of PCB quality (impedance matching), clock skew, and training. At super crazy high 5 000 000 000 bits per second GDDR5 is certainly not CAS3 or much higher than CAS9. The 8 bit prefetch latency is probably big factor. overriding factor.

[SCAVENGER1]

Got my 5870 today, but I can't seem to get the GPUclock utility to work. I'm using the RC7 drivers leaked from MSI, maybe that's the problem...

maybe try the 8.663.1_Beta5_Hemlock_VistaWin7_Nov11 and see if it will help. they are the drivers i am using now on vista and win7 right now.

and so far so good and they run better then the official 9.11 did. since they cause some shader problems in borderlands for me.

but the 9.12 betas fix that for me.

but for me i have not over clocked my 5870 yet since i have no clue of how or what to do.

and from what i have read in this forum i really do not want to since this is my 1st rma 5870.

since my 5870 came to me bad and now this 1 seems to be ok

http://www.filefront.com/15115547/8....Win7_Nov11.exe

[***Deimos***]

Even though this IS xtremesystems, I dont blame you for being cautious with overclocking on expensive new (rma) 5870.
Afterall, its not like you need to overclock to catchup.. its ALREADY the FASTEST.

Besides. 1000Mhz. 1100Mhz. Even 1300Mhz on GPU.. that's peanuts in terms of performance improvement compared to what driver optimizations bring (which also have the advantage of being cumulative and compounded!!).

Judging based on driver timelines of X1900, HD3870, and HD4870 - need HD5xx to wait 3-4 months for most major improvements (DX9/DX10) and probably till summer '10 to top out (especially since virtually no DX11 titles to optimize for right now).

Your HD5870 is only gonna get faster