Discussion in 'Benchmark Mayhem' started by TDurden, Jul 14, 2016.
Not bad at all.
SSE has been with us since at least the PIII days.... His PhenomII has SSE
Yes! Just found it myself actually, but thanks for quick reply.
3DMark's in minimum requires 1.8Ghz DualCore CPU with SSSE3 support, and my Phenom does not have SSSE3 instruction support.
Thanks again tho.
His Phenom II is lacking SSE4 instructions. I've seen quite a few games not running at all becouse that problem so i thought there is a connection.
I have my 280 @ 1120/1400
Whats weird is I have no voltage options in Afterburner even though this is a black edition XFX 280.
Guess I need to make a bios for it.
I used to have a reference Sapphire 7950 boost with a custom bios @ 1300/1500.
Yes its time to snag up a $199 XFX 4GB RX 480 and use it ill high end vega drops.
3898 with my rig all at stock.
Not bad at all.
Your graphics score is 7% faster than i am at stock and SLI disabled, while your CPU is 80% faster lol.
Maxwell cards really benefits from Async enabled :
You can't deny that Maxwell doesn't support Async-Compute as it is stated. If not, how do you explain the performance drop? c:
GTX 970 SLI FTW running at pre-oc speeds. I7 3930k @ 4.3ghz
Graphics Score :7104
I can't remember the exact wording, but think the Oxide dev explained that while Maxwell cards do have support for async compute, they are unable to use it and gain performance, if anything there is a risk of performance loss.
I just tried with and without, and i gained 50 on the graphics score by disabling async compute.
Nearly a year done the line, i'm sure some of us can give up on this every being a feature on Maxwell cards.
Not the end of the world though, performance is still great with this benchmark and competitive considering what i paid.
Just a pity DX12/Vulkan games are not this well optimised.
Not much between our two systems, just my CPU starting to show it's age.
Graphics Score :7335
Indeed. Maxwell cards are good performers overall. It's a shame that wont get any better at this point. Peak of performance and async as you see above, isnt helping at all.
Considering how Kepler cards are performing bad lately (Vulcan and DX12) Maxwell is still standing.
Or just not enabled properly yet. Nv didnt say anything about it with r367 branch, so.. You can continue with your conspiracy theory.
The most this system can get is 7184 so not to shabby!
ok new best!
So what happened there?
3 796 with AMD Radeon R9 290(1x) and AMD FX-8350
Graphics Score 3 924
CPU Score 3 204
Worst motherboard on the planet ASRock 970 Extreme4
GTX980M@1.2 GHz , i74860HQ
Time Spy Result: 3296
Graphics score: 3223
CPU Test: 3790
well nothing changes.
what he said was interesting.
Im not sure of any of it
If you guys made Time Spy and use Async Compute to overlap rendering
, you play right into Pascal's hand. It does not have real parallel execution but
it can fast context switch, with preemption and dynamic load balancing to
improve it's shader utilization if it's under used. At 1080 and 1440p, it is very
much likely not to be using 100% hence the very small gains. People find the
gains drop to almost zero at 4K due to all the shaders being utilized already.
If you guys had actually used a real parallel and multi-engine approach, you would see major gains across the
board for all GPUs capable of this on hardware. Regardless of whether it's a
low shader GPU like 380/X or even the new RX 480, which only have 2304
shaders vs Fury X 4096.