The Test

For the 5700 series launch, AMD issued some new drivers as the previous 8.66 driver set did not include support for these cards. The driver set we used for these cards is 8.66.6, which is from the same branch as the earlier drivers. In our own testing, we haven’t seen any performance differences between these drivers and the previous ones on the 5800 series cards, but AMD did note that certain configurations might see a small performance boost. As such our results are still using the original 8.66 driver for the 4000 and 5800 series.

Also, as AMD sent us a pair of 5770s, we have tested these cards in a Crossfire configuration. This configuration is largely academic, as 2 5770s is just shy of the price of a 5870 and brings with it all of the limitations of multi-GPU scaling as compared to single-GPU scaling.

On a final note, our 5750 sample is a 1GB card.

CPU: Intel Core i7-920 @ 3.33GHz
Motherboard: Intel DX58SO (Intel X58)
Chipset Drivers: Intel 9.1.1.1015 (Intel)
Hard Disk: Intel X25-M SSD (80GB)
Memory: Patriot Viper DDR3-1333 3 x 2GB (7-7-7-20)
Video Cards:

ATI Radeon HD 5870
ATI Radeon HD 5850
ATI Radeon HD 5770
ATI Radeon HD 5750
ATI Radeon HD 4870 X2
ATI Radeon HD 4890
ATI Radeon HD 4870 1GB
ATI Radeon HD 4850
ATI Radeon HD 3870
ATI Radeon HD 4670 512MB
NVIDIA GeForce GTX 295
NVIDIA GeForce GTX 285
NVIDIA GeForce GTX 275
NVIDIA GeForce GTX 260 Core 216
NVIDIA GeForce GTS 250
NVIDIA GeForce 8800GT

Video Drivers:

NVIDIA ForceWare 190.62
ATI Catalyst Beta 8.66
ATI Catalyst Beta 8.66.6
ATI Catalyst 9.9

OS: Windows 7 Ultimate 64-bit

Meet The 5750 Crysis: Warhead
POST A COMMENT

117 Comments

View All Comments

  • hsoverma - Tuesday, October 13, 2009 - link

    I bought a 3870 for about 210.00 a year and a half ago. This card has double the performance, a lot more features, and is starting at 160.00. I figure by christmas, it will be down around 120.00. I run a small frag-box, so the lower heat and lower power makes sense for me over a 4870, and if I ever wanted more power, I could run two of these in crossfire and have all that I need. I am putting this card on my wish-list. Thanks again for a great review. Reply
  • 7Enigma - Tuesday, October 13, 2009 - link

    Great article except for the flaw of not a single Memory OC data point. If as the data shows the 5770 is performing poorly due to inadequate memory bandwidth (seemingly the ONLY issue hampering performance when comparing specs between it and the 4870), it makes sense that a simple OC could shed some light on this issue. Please update the article with some numbers as this card is mainstream enough that I would imagine overclockers *could* see this as a gem in disguise. Reply
  • Ryan Smith - Tuesday, October 13, 2009 - link

    Once things calm down, we're going to do a 5800 series overclocking article. It's something that takes a while to put together, and there are major product launches every week right up through the end of this month. Reply
  • 7Enigma - Wednesday, October 14, 2009 - link

    Can you give us a hint then whether this memory has the potential to overclock well enough to where the bottleneck is overcome? I doubt it can completely remedy the situation but if the memory overclock was enough to make up for the significantly narrower bus in most cases, I think this card would have a better reception. Reply
  • Etern205 - Tuesday, October 13, 2009 - link

    Just a stupid question...

    ATi's Eyefinity requires a display port monitor to work properly (saw on youtube video), and what it does is makes the image share across the screen.

    My question is can you use the ATi card and use all monitors independently without the eyefinity feature?


    Reply
  • squeezee - Tuesday, October 13, 2009 - link

    Yes, you can run the displays independantly just as with any other card. However the 3rd monitor always needs to be Displayport. (or use an ACTIVE displayport adapter) Reply
  • EJ257 - Tuesday, October 13, 2009 - link

    I'm sure there is a perfectly good explanation for this that I'm missing. You say there is 1 SIMD disabled in the 5750 vs the 5770. Looking at the chart on the first page there is a difference of 80 stream processors and 4 texture units between the 5750 & 5770 so this would indicate 4 SIMDs are disabled. Reply
  • EJ257 - Tuesday, October 13, 2009 - link

    Wait, never mind. I see each SIMD consists of 4 texture units and 20 SP (Stream Processors) and each SP contains 4 Stream Cores. I guess in the chart when you say Stream Processors you really mean Stream Cores right? Reply
  • yacoub - Tuesday, October 13, 2009 - link

    Ryan - THANK YOU for including the 8800GT in the graphs: That is the card I (and many other potential buyers) will be upgrading from, moving to a DX11 card.

    It's too bad this new card (5770) can't quite catch the GTX260 C216, as that is its main NVidia competitor in performance and price. It uses just as much Load wattage as a 260, but seems to offer around 80-90% of the performance. Perhaps if they hadn't cut the bandwidth to 128-bit, it would have squashed the GTX-260. But ATI has a habit of under-bussing their cards and it continues to negatively impact high resolution performance, no matter what certain reviewers might claim about the bus-width not hurting performance. Time and again, testing shows potential for improvement from a wider bus.

    Blah. ATI, you always come so close to getting me to purchase but there's always something to hold me back. Perhaps if/when this card drops to $139 (without rebates). But by then NVidia might have their answer out and the GTX-260 would also drop in price or be replaced with a DX11 part, and then the 5770 again loses appeal.
    Reply
  • MadMan007 - Tuesday, October 13, 2009 - link

    I think you'll pass out if you try to hold your breath waiting for a lower midrange NV DX11 answer to these cards. Reply

Log in

Don't have an account? Sign up now