We take the ASUS STRIX GTX 780 OC 6GB video card and run two in SLI and overclock both of these at 4K resolutions to find the ultimate gameplay performance with 6GB of VRAM. We will also compare these to two overclocked ASUS Radeon R9 290X DirectCU II CrossFire video cards for the ultimate VRAM performance showdown.
I read this earlier this morning on Hardocp - Crossfire has taken a lot of bashing over the years and some people or amd haters still think sli is better!
Well this review tells otherwise - Amd, glad your doing a good service for your customers
It would have been better to include an actual FCAT latency reading, that simply relaying on subjective feeling. The tech is there, why not use it to prove your point?
Funny thing is how AMD gets bashed on something that Nvidia does well and they don't but then they catch up and exceed Nvidia.
- Sli used to scale alot better then CFX then the HD6000 series was released with 90% + scalling exceeding the GTX 400 and 500 series in that
- Frame latencies on their single GPU cards were measured to be worse then Nvidias cards but they patched the HD7000 series to exceed the GTX 600 series in that
- Dual GPU frame latencies were considered worse but it seems with the R9 200 series they have exceeded the GTX 700 series in that
Of course it would be better if they could fix these things without them becoming a huge PR issue first but atleast they learn....
Funny thing is how AMD gets bashed on something that Nvidia does well and they don't but then they catch up and exceed Nvidia.
- Sli used to scale alot better then CFX then the HD6000 series was released with 90% + scalling exceeding the GTX 400 and 500 series in that
- Frame latencies on their single GPU cards were measured to be worse then Nvidias cards but they patched the HD7000 series to exceed the GTX 600 series in that
- Dual GPU frame latencies were considered worse but it seems with the R9 200 series they have exceeded the GTX 700 series in that
Of course it would be better if they could fix these things without them becoming a huge PR issue first but atleast they learn....
amd took too long to fix the framelag in crossfire i had 2x 7970 and it was unplayable until by the end of 2013 and other problems with dx9 games, opengl (windows mostly linux)
amd did a good job but they just take some time to release good drivers that was the main reason i switch to nvidia
780 and 780ti are pretty expensive but their quality is superior the heat, overclock, power consumption and better performance in games at release the only problem is the 3gb vram for gaming
Lol, I find it funny that people assume their opinions can invalidate and dismiss objective facts.
Why hasn't Mr. Ryan Shrout not delved deep into his new dvi intercept card's measurements, just after he had announced his prioritised dedication for FCAT, then? Long story short: no one wants to point fingers at the industy's big player.
The games used for the benchmarks apart from Watch Dogs were either AMD optimized or had mantle support. Mantle does provide better performance and great frametimes. Crossfire has improved a lot and it's nice seeing AMD do well. For multi GPU setups though it's better having more memory especially for 4K as games will start getting more VRAM hungry. AMD seems to be in the red though and it would be nice to see some consumer support. Releasing a new flagship GPU to directly compete with the new GTX 880 would be wise, giving them the edge as it will likely be quite more powerful. Although it seems that they might delay it's release for some time giving NVIDIA the upper hand.
I think it's great to see AMD doing so well in price to performance comparisons. It would be nice if they didn't use as much power but not a huge deal breaker. Regarding the whole smoothness thing, why didn't they test it definitively (or try to)? I'm not entirely doubtful that what they're saying is true but it seems common sense to at least try to show the results instead of reverting back to just the reviewer's perception.
Props to AMD though. They're doing a good job of staying competitive with their GPUs.
Lol, I find it funny that people assume their opinions can invalidate and dismiss objective facts.
Why hasn't Mr. Ryan Shrout not delved deep into his new dvi intercept card's measurements, just after he had announced his prioritised dedication for FCAT, then? Long story short: no one wants to point fingers at the industy's big player.
I wouldn't say that "it looks smooth to our eyes" is a real objective fact.
Just average frame rates means nothing today after the whole fiasco of min/max frame times with the AMD cards a year or so ago. It still has a residual bad taste from the past.
AMD did fix the problem in general, but nvidia's SLI smoothness regarding frame times consistency is still one of the best there is and has been proven countless of times. AMD, not as much.
Its pretty simple. FCAT can confirm that AMD frame times is consistent and their drivers are finally fixed and not shabby as they used to be.
Without it, there is left a small doubt in the results. No one said "fake!!" or "no FCAT so its untrue!11!!".
But an added FCAT can be the final nail to confirm this. And its a shame its not there.
The more you overclock, the more you get small inconsistencies and higher ms differences between frames. This can happen to both AMD and Nvidia, and people, including me, would like to see this test done fully.
Does that power consumption number look right? Seems higher than others I've seen (also one of the reasons I've been hesitant to try to SLI/CF higher end cards as I'd be worried I'd have insufficient power).
AMD's software frame pacing has generally worked extremely well, almost from it's inception, in any software where it can actually be enabled.
If it's a DX 10 or 11 game, chances are you aren't going to be able to detect any difference between CFX or SLI in smoothness, even on older hardware that still uses the CFI bridge.
Quote:
It's power draw at the wall with overclocked parts. Hawaii GPUs are extremely power hungry if you can keep them from throttling. I've seen over 330w from a single stock clocked, slightly undervolted, R9 290X in OCCT...even real games can force the card past 200-250w, and this is at the card, not the wall.
Quote:
Originally Posted by B NEGATIVE
You mean the NVIDIA coded FCAT.....
Third party please,using benching softs from one manu to test against another doesn't show objective consideration.
The process takes frames that are tagged with alternating colors and measures their duration. Doesn't matter if the source is NVIDIA hardware, AMD hardware, Intel hardware, or whatever, the FCAT software has no way of knowing.
There is no bias, there is not even any conceivable way to introduce bias with FCAT and hardware frame capture.
Does that power consumption number look right? Seems higher than others I've seen (also one of the reasons I've been hesitant to try to SLI/CF higher end cards as I'd be worried I'd have insufficient power).
Seems perfectly normal, 700 W is about right for high-end SLI. The 290X overclock had maxed out voltage and an insane +50% power cap which definitely increased both power consumption and temperatures a lot over stock, but NVIDIA does have better efficiency both stock and overclocked.
Quote:
Originally Posted by Blameless
AMD's software frame pacing has generally worked extremely well, almost from it's inception, in any software where it can actually be enabled.
If it's a DX 10 or 11 game, chances are you aren't going to be able to detect any difference between CFX or SLI in smoothness, even on older hardware that still uses the CFI bridge.
You're getting pretty annoying with your pedantic replies to say the obvious. Yes, you need a good capture card. That can be pretty special to some reviewers.
Let alone you also need the right methodology, but let's assume you didn't fail to say that.
Status
Not open for further replies.
You have insufficient privileges to reply here.
Related Threads
?
?
?
?
?
Ask a question
Ask a question
Overclock.net
27.8M posts
541.2K members
Since 2004
A forum community dedicated to overclocking enthusiasts and testing the limits of computing. Come join the discussion about computing, builds, collections, displays, models, styles, scales, specifications, reviews, accessories, classifieds, and more!