Hardware Gainward Single PCB GTX 295 Review

True this is a lets test the waters with a single pcb instead of waste a lot of research funds on a newer chip trying to achieve the same thing but then have to recall our new flagship from issues arising, if an issue happens on this card no biggie its already been out on the market for awhile so when they come out with the new dual single pcb gpu card it will look better from a marketing stand point.

Could be driver related but ya that big of a hit does seem a bit odd/.

 

Cool stuff, but i still can't see myself buying this, hell I can't even see myself buying another HD4850.

 

I like the matte black of the card....

 

That's just the early sample we received for review. The retail cards will come with board partner stickers as usual. Sorry if we got your hopes up!

 

Not sure which huge performance hit you're referring too - the HD 4870 X2 seems to take around a 10% performance hit when enabling 4xAA in Crysis or pretty much any other game.

 

Nothing a hairdryer can't fix.

 

Disappointing

 

Beautiful. I want one!

 

it wasnt mentioned but when in quad sli wouldnt the single pcb version in the first pci-e slot have trouble with heat as the fan is dead centre in the middle and so air intake would be covered up via 2nd card while the older 2 pcb has its fans open and at the end of card allowing air to circulate?....if you understand what im saying?

just wondered. cheers.

 

in the folding section you mention that you only could get it to fold on 1 GPU.
forum user JackOfAll is running single PCB GTX295's using both GPU's folding....

Brother of Weapon

 

I havnt a clue how you have come up with the assumption that the 4870x2 is the better and faster card to buy (over all). I have used a gtx295 and it was lets say a round 100. (anything above and below is in percent)... I then decided to sell the gtx295 and get 2 4870x2 cards. Unfortunately one of the cards was dead however the one that wasnt is running as it should. I have had sli gtx260s and yes it is slightly quicker when the 260s are at stock. Anyway I tried the 4870x2 on all games that I play and most that you review. The gtx295 is immensely faster and the numbers you have in your custompc are so horrendously slow for the card its un believable! Have you considered that maybe the gtx295 was faulty or something and wasnt performing as it should or maybe that the drivers you used were just horrific?

In all of my tests the gtx295 was around 15% faster and in some cases up to 30!! I usually take these reviews of yours to be pretty accurate but this is clearly not the case! I now feel I have lost an exceptional investment with the gtx295 after getting 2 4870x2s instead. Im not sure how 2 will perform but I hope to god its better than how 1 does! I should have known better really as your reviews in your mags are so mixed at the moment with the 4870x2 and gtx295 opinions swapping and changing more than my missus's hand bags!

Do you think it would at all be possible to get a hold of another gtx295 and run all the tests again as a strong a claim as it is I am right! The gtx295 is heads above the 4870x2 and I feel slightly robbed!

Crysis according to your latest review runs faster on the 4870x2 than the gtx295 at 1920x1200 with AA in dx10. I have just played a little with the 4870x2 and have gotten clocks a little higher than your 4870x2 to I believe your results to be correct as my system runs a little quicker than yours however the gtx295 was almost double the speed of yours!

Have you thought that maybe the card was only using 1 GPU just like it does in folding at the moment due to an error in drivers? I think need should be addressed as there really is something either horribly wrong with the card you recieved or the card I had was the greatest card ever created. The latter exceptionally doubtful!

Anyway, semi rant over, had a very gutting day after seeing that all I have taken as gospel is actually not true. It even makes me question other reviews you have made.


Andy

 

For the money the 4870 X2 is not a bad investment. The GTX 295 is what.. £100 more expensive? That's not a drop in the ocean.

We image three identical machines with exactly the same installs between cards - we don't even remove drivers, we wipe the whole system. With all due respect, our testing methods have been honed to a T over years of benchmarking. I would never recommend quad-fire. The drivers are so limiting and the depreciation is terrible game to game.

If you could post your setup so we could evaluate the differences in testing procedure and PC drivers it might shed some light as to why you are getting different results.

Personally I think it just goes to show the complete unpredictability of dual GPU cards - we've not been immune to trouble I will admit that, but have you though that maybe your other 4870 X2 is faulty (after all, one of them was) and the GTX 295 is giving abnormally high results in comparison?

 

Evening. The 4870x2 that worked was ever so slightly faster than what yours scored gaing an average of 2fps in almost everything, sometimes 3fps with the average results. My system is an 4.7ghz i7 920 with 12gb of pc1500 (1866) running at a smidge higher than 1866 on an classified.

The hard disc is a 500gb sammy F1.

The gtx295 was ran on a fresh OS (vista 64bit sp2) as was the 4870x2. My gtx295 even clocked in at 732/1588/2286 which is only slightly higher than what you managed with yours but I tested with all at stock.

I even ran vantage and managed a P score in the preset with the 4870x2 of 19k (gpu 17k, over all 19k, cpu 27k). The card was at stock, the system wasnt.

I ran vantage with the gtx295 with identicle settings disabling physx to ensure a fair test and it rocked in at 23.5k (22k gpu, 23.5k over all and 27k on the cpu) These are all only rough numbers but you get the picture.

The biggest supprise was upgrading to the latest driver from nvidia lowered the gtx295 score by almost 1000points but upped crysis warhead by a fair way.

So with ATI finally having decent drivers for crysis and pretty excellent drivers for everything else its finally performing in almost all games as you would hope unlike when it was released. This appears to be the same with this gtx295 from what I can see (awful drivers that you tested with) and as such the card has been shown to be essentially rubbish. The gtx295 from your review was really quite a dissapointment and as I have said, could it have only been running on a single GPU or was the clock speeds limited somehow by the driver?

Im off to the global summit on friday (many thanks there) so maybe I could provide some benchmark results to help my point? I did initially try the card with the provided drivers and it was horrendous in warhead but I thought nothing of it and just installed the newest in the assumption that palit had possibly been stupid enough to provide old drivers with a new card.

Andy

 

A 4.7GHz Core i7 with 1866MHz memory??

That's somewhat significantly faster than we test at - you could simply be highlighting a benefit that a faster CPU makes to each dual GPU graphics card!

I have no idea of 3DMark numbers and we also use 186/9.4 drivers in testing because we have results already we need to compare to (it's simply unfeasible to retest everything every time a new driver drops). We didn't pick the drivers so much - they were the latest available at the time of initial testing.

Palit will just give a CD with the latest drivers at the time of product design. The CDs are stamped en-mass and then it's all shipped out. Never, ever use the graphics driver disk if you can help it

 

ok, for the benefit of making sure all is fair, Ive just lowered my cpu clock and memory clock to match what you have (within reason). The motherboard will make a difference but Im sure its nothing to write home about. I have access to the gtx295 as although sold it hasnt been collected yet. So with this I am going to run the test for crysis warhead again aswell as farcry 2. I shall report back with what I find and what drivers I have used etc. Im not saying your methods of reviewing are rusbbish as they most definitely are not however I do think that there is something not quite right here which I can only assume is driver related.

will edit shortly.

Andy

 

Sure thing, let us know!

 

NOTE: I will edit this post to include gtx295 results shortly.



System i7 920 @ 3.73ghz, 6gb OCZ memory running 1494mhz c9 on classified. Vista SP2 64bit. samsung 500gb F1 (system remains the same for each card)




4870x2 with 9.8 (august o9) drivers at stock 750/900 (750/3600)

Warhead. System preset "enthusiast" (which is all very high if run on crysis) 1920x1200, 4AA 16AF DX10

I ran a short 3minute blast about 45minutes into the game where there was a fair amount of AI and destructible environment 3 times over and used fraps to gain the fps. All of the results were similar so there were no anomalies from what I could see.

Min frame average, 33.4fps
Average frame average, 44.6fps

From what I can see here this is roughly the same as what you have got where maybe the drivers have only changed this by a mere smidge. The slightly quicker ram may have been the culprit but its still safe to say that your results are pretty much the same as mine. However I have noticed that the 4870x2 you have results for says dx9 warhead and gets 32/44. In the next issue you have dx10 enthusiast and get the same so Im guessing a typo of sorts there!



Far Cry 2 DX10.1 1920x1200, 4AA, 16AF

Same scenario here, 3minutes of game play in a busy area. Causing havoc in the middle of town seemed like my best bet.

Min frame Average 47.1fps
Average frame Average 61.0fps

From that I can see here, these are a little slower than your results possibly due to where I tested as this was probably as about as intense as I could find just 25minutes into the game but again its safe to say there within a reasonable window of accuracy compared to yours.






And now the gtx295 single PCB edition.... (made by Palit)

Ive just finished the installation of crysis warhead again on a fresh OS and have had myself a play. The exact same conditions were used with this card, with the AI acting differently the tests are not entirely identical but they are close enough.



So lets begin.

gtx295 single pcb, drivers are 190.62 as the 185.86 that you use for testing your cards dont actually support the gtx295 single PCB (officially). This is probably where you have maybe gone wrong and ended up with the poor results.

Warhead. System preset "enthusiast" (which is all very high if run on crysis) 1920x1200, 4AA 16AF DX10

Min frame average 38.2fps
Average frame Average 49.6fps

(you managed a minimum of 20fps and an average of 25fps which is a far way away from this. This is why I believe your card to have only been running a single core as the results are similar to that of a gtx275 that score 18fps min and 31fps average respectively or that the drivers were crippling performance massively!)

As you can see from the above, the card is actually rather fast and with a premium of around £80-90 for the cheapest model compared to the cheapest 4870x2 cards I would say it is very much worth it!



Far Cry 2 DX10 (as its nvidia) 1920x1200, 4AA, 16AF

Same scenario here, 3minutes of game play in a busy area. Causing havoc in the middle of town seemed like my best bet.

Min frame Average 54.2fps
Average frame average 76.9fps



There results are pretty much on par with what you have previously achieved all be it a little higher.




My conclusion to this is that maybe there was just some sort of conflicting error with crysis warhead that lead to the shockingly slow results. I not saying I am right but it seems to me that there is a possibility that the gtx295 in warhead was running single core effectively producing results similar to that of a slightly underclocked gtx275.

I know that for some reason the gtx295 when used with older drivers only uses 1 of its 2 GPU's when folding and with a driver update it goes back to running both as it should. Maybe this is the case with warhead due to the drivers. I cant be sure but it definitely looks that way.

What do you think??


Andy

 

Finished! I can safely assume the 4870x2 is going to perfom similar to how yours has done even with the newer drivers. I can run Fallout 3, Stalker clear sky and a few other games if required. I just think that maybe this card should be re-reviewed maybe where it failed with newer drivers??


Andy

 

Will it Fold?

No mention of folding in the review but a suspiciously low (similar to GTX275) score.

I tried to run a folding rig with two of these, no joy, or rather only two GPUs. Anyone else tried it?

 

I can find out for you but you will have to wait a few days lol!

Andy