GTX 465 Golden Edition.
wish i still had my 6450 so that i could torture it with this test
only got a mild bump up... was artifacting like a mofo once it hit 70c. thought it was gonna crash by the time it hit 77c.
When you get this thing under water it's going to crush my score. XD
That said, I did a little breakdown on really what it took to beat a 980 Ti and Titan X alike with essentially four 6970s.
Combined stats without CFX scaling counted:
GPixel/s: 123
GTexel/s: 369
A rough estimate for scaling on the 6900 series is around 93% at its best. If that stays the same across the entire lineup, we get the following stats.
GPixel/s: 114.4
GTexel/s: 343.2
And finally for the worst case scaling scenario, 90% across the first two cards, then 82% across the last two cards.
GPixel/s: 105.7
GTexel/s: 299.2
That last scenario looks a lot closer to the stats of your 980 Ti. The only outlier is the memory speed, which I think just comes down to the way the benchmark is setup here. Memory speed and capacity isn't important for 1080p with very little AA, so 2GB at 179GB/s won't perform much less than 6GB at 379GB/s.
I think its more due to the very poor scaling in this benchmark with multi gpu.
3 of those cards should run circles around a 980ti
Makes sense.
Once I ran witcher 3 at 1540mhz for lulz. the artifacting was absurd but it ran for a few minutes. I'll probably be getting a bitspower block as soon as I have the money.
Eh, no not really. This is a 40nm process with little more than half as many SPUs, built on the salvaged remains of the VLIW architecture and limited to a specific draw call amount. The 980 Ti holds every advantage, and so the fact that even four 6970s can compete and win is astonishing. If AMD had got the 1792 SPU 32nm chips to work then sure, but I don't think any of those even made it onto a PCI-E card for final testing.
Thanks TSMC for that BS. They had a contract and decided to stop development after setbacks, forcing AMD back to 40nm. Cayman was supposed to be a significant improvement...
Everything at stock.
Part of me is tempted to grab my old 8350 just to compare how bad the bottleneck I had was.
this benchmark wouldn't show it any bottleneck even if it was an intel atom.
On max settings it will. It showed a meager 8% bottleneck when I compared the 5820K to the 8350. That whole deal was NOT worth the upgrade.
Actually I just looked at the old scores and it's not even 8%
FX-8350 at 5.1GHz holds up to the i7-5820K at 4.4GHz. At least in a gaming sense.
Shame, some games I had with the 8350 were simply unplayable. But I suppose makes sense since it's a gpu benchmark not a gaming benchmark.
The main issue in gaming is, that you had 8 cores and the games max out 2 of them and the rest are picking their noses... Not that it's great cpu, but the games don't even use it properly...
Benchmarks like these don't use cpu at all.
I'm sure you could get away with a core 2 quad running at stock without issues.
They DO use the CPU a bit. The Turion 64 X2 bottlenecks so much that even the 6990 can only score 900 points. While a Core 2 Quad wouldn't bottleneck that badly, it would still affect the benchmark a little.
Why would you test it on a ancient low power mobile chip :D?
900 points is still impressive for something about as powerful as a phone arm chip.
Because I have a mini-ITX Socket S1G1 motherboard that has a PCI-E X16 slot on it. :P
1 Like
At 743MHz it killed my power.
Also, GPU-Z has got to be having a laugh with this reading. 32 x 0.722 = 23.1 not 10.1... I don't remember why GPU-Z can't read Fermi properly.
try running your CF setup using ARF Friendly (you can create new profile for application -- remember you need to select proper exe file as when you run heaven its not the same exe file you run from shortcut).
you'll see your performance jump like hell.