• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Radeon R9 295X2 In CrossFire review

Kaapstad

Kaapstad

Kaapstad

Man of Honour
Joined
21 May 2012
Posts
31,943
Location
Dalek flagship
http://www.tomshardware.com/reviews/radeon-r9-295x2-crossfire-performance,3808.html#xtor=RSS-182

Before anyone says it I do think the above is a pretty useless review as their methods need a lot of improving.

What is interesting is how they got on with Tomb Raider and frame pacing, I have seen this myself and it is pretty awful with 4 Hawaii chips. I have also noticed that the performance on TR with the latest drivers (14.4) is very much lower than 13.11 and I suspect that AMD have done this to try and improve the situation with TR frame pacing. Visually TR on 14.4 is a lot better than 13.11 but it is still very poor.
 
Kaapstad said:
http://www.tomshardware.com/reviews/radeon-r9-295x2-crossfire-performance,3808.html#xtor=RSS-182

Before anyone says it I do think the above is a pretty useless review as their methods need a lot of improving.

What is interesting is how they got on with Tomb Raider and frame pacing, I have seen this myself and it is pretty awful with 4 Hawaii chips. I have also noticed that the performance on TR with the latest drivers (14.4) is very much lower than 13.11 and I suspect that AMD have done this to try and improve the situation with TR frame pacing. Visually TR on 14.4 is a lot better than 13.11 but it is still very poor.
Click to expand...

Interesting.

What i gathered from that review is 295x2 is without doubt faster than 780TI SLI.

Why didn't they use Mantle for BF4? The 3/4 card scaling on it is superb, much better than DX scaling. Not to mention the frame times are way better than DX, though AMD still excel at frame times in BF4 in DX anyway.

Looks like four way TR still needs some work though on the frame variance.
 
The Titan SLI results pretty much compare with what I get with everything on stock, so good to see Tom's being fair on the Titans at least.
 
Gregster said:
The Titan SLI results pretty much compare with what I get with everything on stock, so good to see Tom's being fair on the Titans at least.
Click to expand...

4 Titan's would have annihilated the two 295s on Crysis 3 as what that review does not show is the drivers don't work for the AMD 4 GPU setups on Crysis 3. Edit, wrong review, I was looking at another review today that was even worse.

Tom's seem to have found some of the worst games to use for 4 cards setups too, MLL is useless with both AMD and NVidia for example. Having said that though 4 Titans on TR are very smooth indeed.

Another problem with reviewing 4 GPUs is they really must use the absolute max settings or you will get problems with negative scaling, running the CPU at 4.2 was a big mistake and using a 1350 watt PSU:D was probably another.
 
Last edited:
Small gains indeed for running quadfire it seems a lot of money for little performce :(

Nice cards and good to see AMD try a offbeat way of cooling imo
 
TheRealDeal said:
This has some 4 way scaling with a lot more success. The only game they are using that's the same is battlefield with much better scaling and uses dx and mantle.

http://www.digitalstormonline.com/unlocked/4-way-quad-crossfire-amd-r9-295x2-benchmarks-at-4k-idnum228/
Click to expand...

Good find.

Kaap, looks like you might have a case of PEBKAC. :D

gWnMI1d.jpg


Jz0nVzi.jpg


xtqTWA0.jpg


39etbrd.jpg


Qg8XLav.jpg
 
In case Kaap wants to do some tests he was asked where he tested.

WHICH LEVEL OF CRYSIS 3 you have tested i have this cards i would like to test to ?
Click to expand...

It is the second level after you get the alien sniper and take out the three guns on the tower. After you meet up with the other character and jump into the sewer. If you watch our video, it's that sequence.
Click to expand...
 
Last edited:
TheRealDeal said:
In case Kaap wants to do some tests he was asked where he tested?
Click to expand...

Just retested Crysis 3 and it looks like they have finally fixed quadfire in the game.

When I tested Crysis 3 last night I was only getting about 15fps, this may have had something to do with running BF4 on mantle just before that.:D

What I said about TR still stands though, it is a mess with 4 Hawaii GPUs. Having said that it is better than it was on13.11.
 
Kaapstad said:
Just retested Crysis 3 and it looks like they have finally fixed quadfire in the game.

When I tested Crysis 3 last night I was only getting about 15fps, this may have had something to do with running BF4 on mantle just before that.:D

What I said about TR still stands though, it is a mess with 4 Hawaii GPUs. Having said that it is better than it was on13.11.
Click to expand...

Digitalstorm posted Quad Fire 290X benches for Crysis 3 back in November with similar fps and scaling Kappers.

I will forward on your findings about TR though. :)
 
Last edited:
TheRealDeal said:
This shows scaling at 4k on 1,2,3,4 way systems not long after launch.

To be honest scaling on the amd side look's every bit as good as NV's back then.

http://forums.anandtech.com/showthread.php?t=2359120

Lot's of games tested
Click to expand...

I would have liked to have seen them use max settings in all the games but their Crysis 3 tests do show very little gained going from 3 to 4 cards which is what I found on the 13.xx drivers. If they were to rerun it now with 14.4 they would be hitting about 60fps instead of the 40fps they got.
 
Kaapstad said:
I would have liked to have seen them use max settings in all the games but their Crysis 3 tests do show very little gained going from 3 to 4 cards which is what I found on the 13.xx drivers. If they were to rerun it now with 14.4 they would be hitting about 60fps instead of the 40fps they got.
Click to expand...

It does show though in capable hands unlike toms the majority of games are scaling and it's not like they only used a few games. Then again tom's was using 2 x 295 and not 4 x 290x. I would imagine there should not be to much of a difference though. When working i would expect the 2x295 to give better results as i suspect there would be a decent amount of throttling using 4 x stock cooled 290x's on a test bed.
 
Last edited:
TheRealDeal said:
It does show though in capable hands unlike toms the majority of games are scaling and it's not like they only used a few games.
Click to expand...

+1

I don't think Tom's have got a clue when it comes to benching 4 way setups, sadly the same applies to a lot of the review sites. A lot of these sites will use the same setup they have used for single and 2 way reviews and this just does not work.
 
TheRealDeal said:
Then again tom's was using 2 x 295 and not 4 x 290x. I would imagine there should not be to much of a difference though. When working i would expect the 2x295 to give better results as i suspect there would be a decent amount of throttling using 4 x stock cooled 290x's on a test bed.
Click to expand...

Going on what Gibbo posted the other day and comparing to 4 single Hawaii GPUs, clock for clock the performance is almost indentical.

On an open test bed 4 singles on air will beat a couple of 295s but you will need ear plugs when overclocking.:D In a case 4 singles can get a bit warm on air though and is something that should be avoided.

The biggest problem with 4 singles on air is the heat generated, you can keep the cards below 94C without too much trouble with a high fan profile but all that heat energy seems to effect the rest of the system and cause instability. The first thing I noticed after fitting waterblocks to the GPUs was the rest of the system was a lot more stable.
 
Kaapstad said:
Going on what Gibbo posted the other day and comparing to 4 single Hawaii GPUs, clock for clock the performance is almost indentical.

On an open test bed 4 singles on air will beat a couple of 295s but you will need ear plugs when overclocking.:D In a case 4 singles can get a bit warm on air though and is something that should be avoided.

The biggest problem with 4 singles on air is the heat generated, you can keep the cards below 94C without too much trouble with a high fan profile but all that heat energy seems to effect the rest of the system and cause instability. The first thing I noticed after fitting waterblocks to the GPUs was the rest of the system was a lot more stable.
Click to expand...

Yea i think the review sites test with stock fan profiles so my thinking was the watercooled 295's would be fine but the 4 290x's would suffer more.
 
You must log in or register to reply here.
Back
Top Bottom