[H] AMD Radeon R9 295X2 CrossFire Review

[toyota]

its ridiculous that out of only 4 games, 1 of them could not even work correctly with that setup.

 

[Lepton87]

its ridiculous that out of only 4 games, 1 of them could not even work correctly with that setup.

If you have used Quad GPU configurations before you would know that it actually worked extremely well, often offering similar scaling from 2 to 4 gpus as from to 1 to 2 gpus. Do you think Quad GPU from NV works any better? Even NV admitted that it doesn't in a roundabout way. And there is nothing ridiculous about it, it's ridiculous to draw any conclusions from such a small sample size, it shows extreme bias.

 

[toyota]

If you have used Quad GPU configurations before you would know that it actually worked extremely well, often offering similar scaling from 2 to 4 gpus as from to 1 to 2 gpus. Do you think Quad GPU from NV works any better? Even NV admitted that it doesn't in a roundabout way.

do you even know what I am referring to? Eyefinity nor 4k would work in Far Cry 3 with that setup.

 

[caswow]

If you have used Quad GPU configurations before you would know that it actually worked extremely well, often offering similar scaling from 2 to 4 gpus as from to 1 to 2 gpus. Do you think Quad GPU from NV works any better? Even NV admitted that it doesn't in a roundabout way. And there is nothing ridiculous about it, it's ridiculous to draw any conclusions from such a small sample size, it shows extreme bias.

NV didnt want to be a contender in this review

We are planning a R9 295X2 QuadFire review. Would NVIDIA like to have a contender in this review? If so, let me know what you might have.

The reply we got back from Bryan Del Rizzo was:

No Thanks.

:awe:

 

[Skurge]

do you even know what I am referring to? Eyefinity nor 4k would work in Far Cry 3 with that setup.

I'm pretty sure he know what you are talking about. and I agree with him.

 

[toyota]

I'm pretty sure he know what you are talking about. and I agree with him.

agree with him about what? what is there to argue about? its a fact that the setup did not work for them with Far Cry 3 in Eyefinity or 4k.

 

[skipsneeky2]

Nice to see 4k is actually playable in BF4 with max settings, a average of 74fps with 2 295x2 honestly isn't all that bad but the 2x AA numbers look a lot more tasty with 89fps.

A generation or two is probably all we need to see a pair of single $500 or lower cards in sli or crossfire to best those numbers given by 295x in crossfire but still nice to see it can happen if you got the cash to spend today.:biggrin:

 

[3DVagabond]

So, it uses more power and performs better? If that's going to be the only thing people have to complain about, that's not too bad. It's a halo part. Let's not forget that.

 

[Skurge]

agree with him about what? what is there to argue about? its a fact that the setup did not work for them with Far Cry 3 in Eyefinity or 4k.

Nobody is arguing about if Far Cry works or not. I didn't. I agree with him when he said it's not too bad considering Quad GPU setups always have a bunch of issues. So for them to only have an issue in 1 game is not to bad. It is certainly not ridiculous. Looking at other tech sites that have done Quad GPU testing with 295x2s. It seems to work well.

 

[toyota]

Nobody is arguing about if Far Cry works or not. I didn't. I agree with him when he said it's not too bad considering Quad GPU setups always have a bunch of issues. So for them to only have an issue in 1 game is not to bad. It is certainly not ridiculous. Looking at other tech sites that have done Quad GPU testing with 295x2s. It seems to work well.

well sorry but I disagree then when you only test four games and one of them doesnt even work. but yeah I guess I am crazy for expecting things to work after spending 3000 bucks on gpus.

 

[Lepton87]

do you even know what I am referring to? Eyefinity nor 4k would work in Far Cry 3 with that setup.

FarCry3 isn't an AFR friendly game, when I had 4 Caymans in my rig, I had to play it on just one. The point is 4 games is such a small sample size that you can't really draw any conclusions from it. When one game out of fuor didn't work you can't extrapolate that you will have problems in 25% of games the sample size is just too small. It literally tells nothing, it may have been just bad luck or you might as well got lucky and much more than 25% of games might not work.

well sorry but I disagree then when you only test four games and one of them doesnt even work. but yeah I guess I am crazy for expecting things to work after spending 3000 bucks on gpus.

It seems that you have no idea what a sample size is or how statistics work. And yes, you are crazy to expect it to work EVERY TIME properly. If you had any prior experience with such set-ups you would know how crazy it is to expect it to work every time in every game. It's something so niche that maybe 0.01% of gamers use four gpus. If expecting a company to perfect an offering that is so extremely niche isn't crazy I don't know what is.

 

[toyota]

FarCry3 isn't an AFR friendly game, when I had 4 Caymans in my rig, I had to play it on just one. The point is 4 games is such a small sample size that you can't really draw any conclusions from it. When one game out of fuor didn't work you can't extrapolate that you will have problems in 25% of games the sample size is just too small. It literally tells nothing, it may have been just bad luck or you might as well got lucky and much more than 25% of games might not work.



It seems that you have no idea what a sample size is or how statistics work.

lol did I say anything about 25% of all games not working just because of that? and I made an A in that class so I dont need a remedial explanation. I dont give a damn about actual statistics here and just care about every game working so to me its horrible when you only test 4 games and 1 of them cant even work with the 3000 bucks in gpus you just bought.

 

[RaulF]

Which is funny to say given geforce cards boost to 1200MHz by itself. Kind of hypocrisy

Some do, only if their base clock is higher than factory via bios.

But no, not all geforce GK110 boost to 1200, and not all are stable at that speed.

 

[Lepton87]

lol did I say anything about 25% of all games not working just because of that? and I made an A in that class so I dont need a remedial explanation. I dont give a damn about statistics and just care about every game working so to me its horrible when you only test 4 games and 1 of them cant even work with the 3000 bucks in gpus you just bought.

Keep dreaming that every game will just work on something so niche, regardless if it is from NV or AMD. It doesn't matter that you didn't say anything about 25% of games, you focused on that one game as if it meant anything, IT DIDN'T. Such set-ups aren't problem free, you can't expect every game to just work, just keep to one GPU if you want that.

 

[toyota]

Keep dreaming that every game will just work on something so niche, regardless if it is from NV or AMD. It doesn't matter that you didn't say anything about 25% of games, you focused on that one game as if it meant anything, IT DIDN'T.

you are right. that one game should mean nothing after spending 3000 bucks. just go play the ones that work.

 

[Silverforce11]

It is interesting that Mantle doesn't work with 4 GPUs, scaling really bad in MP.

PCPER also did QuadFire test but they did Singleplayer and got excellent scaling. So there's room for improvement via either BF4 patching or drivers to boost MP scaling.

@toyota
When you are gaming on the edge, there's always issues, for both vendors. Recall on launch, 4K on several games (go read the release [H] review) didn't even work with SLI they had to issue a driver fix.

I'm sure people who make the purchasing decision to go with Quad Fire/SLI know full well which games they are spending their time on.

 

[Skurge]

well sorry but I disagree then when you only test four games and one of them doesnt even work. but yeah I guess I am crazy for expecting things to work after spending 3000 bucks on gpus.

HardOCP had a problem in 1 game. PCPER had no problems in any of their games, PCGameshardware had no problems.

You can't draw conclusion from any games. Also no matter how much you spend there will always be issues. Some of it will be hardware, some of it will be drivers, some of it will be the games. It's just something you will have to accept as a PC gamer.

 

[Skurge]

It is interesting that Mantle doesn't work with 4 GPUs, scaling really bad in MP.

PCPER also did QuadFire test but they did Singleplayer and got excellent scaling. So there's room for improvement via either BF4 patching or drivers to boost MP scaling.

@toyota
When you are gaming on the edge, there's always issues, for both vendors. Recall on launch, 4K on several games (go read the release [H] review) didn't even work with SLI they had to issue a driver fix.

I'm sure people who make the purchasing decision to go with Quad Fire/SLI know full well which games they are spending their time on.

SLI still doesn't work on none-MST 4k monitors at the moment. There should be a driver out soon to fix that, but that is an example. 0% of games would work with that setup, no matter how much money you spent on GPUs.

 

[3DVagabond]

There are always games that don't work/scale in multi GPU. It's the nature of the beast. It can require support from both the hardware and software vendors. Sometimes they don't share the required info. Sometimes it's just not at the top of the priority list. I feel for inexperienced users who don't realize this. Anyone I've seen in this thread I believe is fully aware.

Toyota, I know you realize this. I'm not sure why you are making such a big deal. Especially, the "one game out of four" aspect like that has any relevance at all. If [H] wanted to they could have easily selected four games that all worked, or they could have selected four games that didn't. I believe (this is just me) that there might have been one game that didn't work in the mix to show people that multi GPU isn't perfect. It would have been less of a service to their readers to have done it any other way. Again, IMO.

 

[toyota]

There are always games that don't work/scale in multi GPU. It's the nature of the beast. It can require support from both the hardware and software vendors. Sometimes they don't share the required info. Sometimes it's just not at the top of the priority list. I feel for inexperienced users who don't realize this. Anyone I've seen in this thread I believe is fully aware.

Toyota, I know you realize this. I'm not sure why you are making such a big deal. Especially, the "one game out of four" aspect like that has any relevance at all. If [H] wanted to they could have easily selected four games that all worked, or they could have selected four games that didn't. I believe (this is just me) that there might have been one game that didn't work in the mix to show people that multi GPU isn't perfect. It would have been less of a service to their readers to have done it any other way. Again, IMO.

why is it so hard to comprehend that if you test only 4 games and 1 of them does not work that it looks worse than testing 50 and only 1 not working?

 

[3DVagabond]

why is it so hard to comprehend that if you test only 4 games and 1 of them does not work that it looks worse than testing 50 and only 1 not working?

Because the sampling is too small to be statistically meaningful.

 

[toyota]

Because the sampling is too small to be statistically meaningful.

if you bought 4 of something and 1 of them did not work would you give a rat's behind about sample size at that moment? again this is not about statistics. I am saying it "appears" worse when you have a game not work when you are only testing 4 of them.

 

[3DVagabond]

if you bought 4 of something and 1 of them did not work would you give a rat's behind about sample size at that moment? again this is not about statistics. I am saying it "appears" worse when you have a game not work when you are only testing 4 of them.

It only appears that way until you think about it. With a sampling size of 4 it's not enough to make an informed decision on the overall percentage of games that will or won't run with 4 GPU's.

In the end, that wasn't the purpose of this review. You would have to test a lot of games and then decipher any trends to hopefully have any meaningful info on what games you might expect to run. You might, for example, be able to determine that games in DX# show a trend of not running. You might be able to determine that games on engine "X" tend not to run. You might be able to determine that TWIMTBP games tend not to run. You might be able to determine that Only GE games give support. We might find it's an issue with the mobo and/or it's BIOS. Or the chipset. Etc...

The only thing we can take from this with any certainty is that FC3 wasn't stable and crashed @ the 2 resolutions they tested with Quadfire on their system. The best we can do is check the multiple reviews that are out there and see if there is a trend. Even then, we don't have enough concrete info, but we could draw a more meaningful conclusion that would likely be more accurate overall.

 

[Enigmoid]

Things I don't like about the article.

1. Four games is much too small a sample size to draw any conclusions making the review somewhat pointless as a general performance guide.

2. Article never really mentions that a lot of games are NOT going to scale onto 4 gpus good drivers or bad drivers.

3. No 780 Ti crossfire results or 780 ti quad sli numbers to compare to or quad crossfire R290X. Why are there power numbers for 780Ti SLI but not performance numbers?

4. The penmanship is poor. I don't know what is with HardOcp but they need a better editor/writer.

Example

Other games performed well. We saw big scaling in Crysis 3 and Far Cry 3 and Tomb Raider. We also found that game settings can affect the scaling. In very GPU intensive settings, and on displays that are GPU intensive like 4K the scaling is higher. Therefore if you want the best out of QuadFire the idea is crank up the settings as high as you can to get the performance you paid for.

Its awkwardly worded and while irrelevant to the method and results detracts from the review.

 

[Silverforce11]

Things I don't like about the article.

1. Four games is much too small a sample size to draw any conclusions making the review somewhat pointless as a general performance guide.

2. Article never really mentions that a lot of games are NOT going to scale onto 4 gpus good drivers or bad drivers.

3. No 780 Ti crossfire results or 780 ti quad sli numbers to compare to or quad crossfire R290X. Why are there power numbers for 780Ti SLI but not performance numbers?

4. The penmanship is poor. I don't know what is with HardOcp but they need a better editor/writer.

Example

Its awkwardly worded and while irrelevant to the method and results detracts from the review.

1. [H] only test recent GPU intensive games, or "popular" ones. They have always had a small list of games. But thats because they do proper testing, ie, long play through and not canned benches and for BF4, always 64MP! I don't know of many other good sites that review BF3/4 with a focus on its actual targeted play mode: MP.

2. Looking at the review from PCPER as well as Kitguru, it does seem like a LOT of games do actually scale very well with 4 GPU, much better than previous generations. The only outlier is Bioshock (already 100+ fps) and Metro LL and BF4 in MP (SP it scales like a champ).

3. It wouldn't be a contest since at 4K, R295X2 already destroys 780ti SLI based on the games [H] test at, with a performance deficit of 20-30%. Seeing how well Hawaii scales with multi-GPU, it would be a very ugly contest for Quad 780ti, which is why when [H] asked NV for a contender, they refused. Why make it look much worse, right?

4. That's their pen-style, hate it or love it. I'm neutral, I look at the data and gameplay summary.

At the end of the day, [H] is only relevant for the games you are interested in, for for an overall performance comparison, they are not the go-to review site due to their small sample of games. Saying that, their sample is often the most popular at the time and/or most intensive, so it does have some weight.