Which one of these GPU's would you choose and why?

[soarwitheagles]

Hi again!

I recently acquired a new XFX Radeon HD 5770 1GB card.  I am presently using Galaxy's GTX 460 768MB card in both rigs.

Here's my questions:

1. Which card is the better card [Tom's Hardware GPU chart is showing both cards at the same tier]?

2. Would the ATI card work better in the AMD rig?

3. Would the Nvidia card work better in the Intel rig?

Here's the specs and info on each card:

XFX HD-577A-ZNFC Radeon HD 5770 (Juniper XT) 1GB 128-bit GDDR5 PCI Express 2.1 x16 HDCP Ready CrossFireX Support Video Card:

http://www.newegg.com/Product/Product.aspx?Item=N82E16814150447

Model

Brand
    XFX

Model
    HD-577A-ZNFC

Interface

Interface
    PCI Express 2.1 x16

Chipset

Chipset Manufacturer
    ATI

GPU
    Radeon HD 5770 (Juniper XT)

Core Clock
    850MHz

Stream Processors
    800 Stream Processing Units

Memory

Effective Memory Clock
    1200MHz (4.8Gbps)

Memory Size
    1GB

Memory Interface
    128-bit

Memory Type
    GDDR5

3D API

DirectX
    DirectX 11

Ports

HDMI
    1 x HDMI

DisplayPort
    1 x DisplayPort

DVI
    2 x DVI

General

RAMDAC
    400 MHz

Max Resolution
    2560 x 1600

CrossFireX Support
    Yes

Cooler
    With Fan

System Requirements
    Minimum 450 Watt Power Supply Requirement

Power Connector
    6 Pin

Dual-Link DVI Supported
    Yes

HDCP Ready
    Yes

Card Dimensions
    8.5" x 4.38" x 1.5"

Features

Features
    ATI Eyefinity Technology


And here are the specs for the Galaxy GTX 460:

Galaxy 60XMH6HS3HMW GeForce GTX 460 (Fermi) GC 768MB 192-bit GDDR5 PCI Express 2.0 x16 HDCP Ready SLI Support Video Card:

http://www.newegg.com/Product/Product.aspx?Item=N82E16814162058&cm_re=galaxy_gtx_460-_-14-162-058-_-Product

Model

Brand
    Galaxy

Model
    60XMH6HS3HMW

Interface

Interface
    PCI Express 2.0 x16

Chipset

Chipset Manufacturer
    NVIDIA

GPU
    GeForce GTX 460 (Fermi)

Core Clock
    700MHz

Shader Clock
    1400MHz

Stream Processors
    336 Processor Cores

Memory

Effective Memory Clock
    3696MHz

Memory Size
    768MB

Memory Interface
    192-bit

Memory Type
    GDDR5

3D API

DirectX
    DirectX 11

Ports

HDMI
    1 x mini HDMI

DVI
    2 x DVI

General

Max Resolution
    2560 x 1600

SLI Support
    SLI Ready

Cooler
    With Fan

System Requirements
    Required 450W power supply

Dual-Link DVI Supported
    Yes

HDCP Ready
    Yes

Features

Features
    NVIDIA CUDA technology with CUDA C/C++, DirectCompute 5.0 and OpenCL support
    NVIDIA PhysX technology
    NVIDIA 3D Vision Surround Ready
    NVIDIA PureVideo HD technology
    OpenGL 4.0 support

[Dark Mantis]

Hi Soar,

Well not so much from a scientific point of view but more a statistical one I would go for the AMD Radeon. You only have to check around and see where most of the problems with GPUs come from.

[soarwitheagles]

Hi Soar,

Well not so much from a scientific point of view but more a statistical one I would go for the AMD Radeon. You only have to check around and see where most of the problems with GPUs come from.

DM,

Thanks for your input!  Wow, I never would have thought you would recommend the ATI over Nvidia.  I have never had a problem with Nvidia.  And I have never used ATI on a long term consistent basis before.

Thanks for your input!

Soar

[Dark Mantis]

That's it Soar I just call it as I see it. I have installed Radeons in my builds for some time now and not had any real problems at all. I know a lot of people could say the same about nVidia but just from the issues that I see arrising  all the time most seem to be nVidia based.

[soarwitheagles]

DM,

Ok, perhaps I will give the 5770 a whirl!  As I mentioned before, I have never used ATI on a long term basis [only installed it for testing purposes in the past].

I am not sure which of these two will work best....

460 GTX has 768 MB compared to the XFX 5770's 1 full GB.
460 GTX has 336 stream processors compared to the 5570 800 processors!
460 GTX has 192-bit compared to the 5570's 128-bit

Are these factors important in choosing a GPU?

Thanks,

Soar

[Wonderwrench]

DM,

Ok, perhaps I will give the 5770 a whirl!  As I mentioned before, I have never used ATI on a long term basis [only installed it for testing purposes in the past].

I am not sure which of these two will work best....

460 GTX has 768 MB compared to the XFX 5770's 1 full GB.
460 GTX has 336 stream processors compared to the 5570 800 processors!
460 GTX has 192-bit compared to the 5570's 128-bit

Are these factors important in choosing a GPU?

Thanks,

Soar

The cards should perform about the same though the GTX 460 768 meg card will be the faster of the two in general. The only time this will not hold true is if a game requires more video memory than the GTX 460 has or the game in question just plays better on ATI hardware. The chances that the HD 5770 will be faster increases as the resolution increases. If you run 1920x1080 res when gaming and you play the latest games neither card will will perform acceptably without lowering in game graphics settings. Cards in this class are good performers at resolutions of 1680x1050 or less.

Bill

[soarwitheagles]

Bill,

Wow!  You know GPU cards!  Ok, thank you for the detailed info....that was exactly what I was looking for!

Have a wonderful day Bill!

Soar

[Badbonji]

The HD5770 is equivalent to a GTS450 in terms of performance. You cannot directly compare shaders either between Nvidia and ATI as they work in different ways. ATI's shaders work in groups of 5.

As for Nvidia vs ATi, I have had bad luck with both. Recently a GTX260 has died (awaiting the oven treatment!) and a HD4870 has died. Right now my desktop won't boot, all I wanted to do was play Black Ops seeing as my last exam was today! Not sure what the problem is there as of yet, could be the HD5970 or might be PSU etc.

[Dark Mantis]

You certainly seem to be getting your share of bad luck as far as the GPUs goes.

Are any of the faults common in any way ?

[soarwitheagles]

The HD5770 is equivalent to a GTS450 in terms of performance. You cannot directly compare shaders either between Nvidia and ATI as they work in different ways. ATI's shaders work in groups of 5.

As for Nvidia vs ATi, I have had bad luck with both. Recently a GTX260 has died (awaiting the oven treatment!) and a HD4870 has died. Right now my desktop won't boot, all I wanted to do was play Black Ops seeing as my last exam was today! Not sure what the problem is there as of yet, could be the HD5970 or might be PSU etc.

BB,

So sorry to hear about your GPU's!  I hope you can resolve the problem swiftly!

The benchmarks I have read on the HD5770 and the GTS450 reveal a substantial performance difference...[please see benchmark chart].

Perhaps you meant the 460?

Either way, once I looked at that benchmark chart, I think I should stick with the GTX 460's [and one part of me cried for selling off both my brand new 480's]!  The 460's I presently use appear to be perform better than the 5770's.  I didn't see the chart earlier...now I book marked that benchmark url because it is updated daily and I think it is fairly accurate....how about you...do you sense it is accurate or are they way off?

http://www.videocardbenchmark.net/high_end_gpus.html

Soar

PS Congrats on your last exam!  Sure feels good doesn't it?

[Badbonji]

Yeah, until next year anyway! Still have 2 years of uni left before I graduate.

My main desktop will not boot after sitting idle for a few days, then will suddenly work again without changing any components. Think it must be the PSU or motherboard...

The GTS450 is an updated GTX260 like the HD5770 is to the HD4870. Is that benchmark passmark? That one is very unreliable, it shows cards all over the place. Dual GPUs are far lower than they should be, and it shows the HD5870 slower than the HD5770!

In a game, the 450 is very close to the 5770:

http://www.bit-tech.net/hardware/graphics/2010/09/13/nvidia-geforce-gts-450-review/5

[soarwitheagles]

BB,

Yeah, that was passmark.  I thought it was reliable!  Well, perhaps I should go back to using the Tom's Hardware GPU hierarchy charts!

I do believe both charts show the 460's are the best of the 3.  I suppose I will Craigslist the 450, 4850, and the 5770 and keep the two 460's I have now.

Hey, sorry about your main rig not being able to boot!  Have you a PSU tester?  I purchased a PSU tester and it works wonders for me in finding obvious voltage discrepancies.  For $15 it has been a good investment for me.

What university do you attend and what is your major if I may ask?

I've had two majors: Kinesiology and Child Development.  Each university experience was good and positive!

Here in the USA, Fry's Electronics will test our Mobo's for free if we purchased them there.  I have no clue how to test Mobo's.  You may want to ask DM more on that!

Have a good one!

Soar

[Wonderwrench]

BB,

Yeah, that was passmark.  I thought it was reliable!  Well, perhaps I should go back to using the Tom's Hardware GPU hierarchy charts!

I do believe both charts show the 460's are the best of the 3.  I suppose I will Craigslist the 450, 4850, and the 5770 and keep the two 460's I have now.

Hey, sorry about your main rig not being able to boot!  Have you a PSU tester?  I purchased a PSU tester and it works wonders for me in finding obvious voltage discrepancies.  For $15 it has been a good investment for me.

What university do you attend and what is your major if I may ask?

I've had two majors: Kinesiology and Child Development.  Each university experience was good and positive!

Here in the USA, Fry's Electronics will test our Mobo's for free if we purchased them there.  I have no clue how to test Mobo's.  You may want to ask DM more on that!

Have a good one!

Soar

I would not call Passmark PerformanceTest unreliable though the GPU 3d test needs some knowledge to interpret it correctly. First the 3d test can't test SLI or CF. This is not the fault of the program. Nvidia and AMD/ATI need to add support to their drivers. Passmark has contacted both parties several times over the years and neither party bothered to respond. The lack of SLI/CF support means you can't test or compare the performance of SLI/CF systems. Dual GPU cards will show lower performance than the single cards they were derived from because most dual GPU cards run the GPU's at lower clock speeds. With that out of the way. The 3d test has several tests in windowed mode and several tests in full screen but non HD resolutions. This can cause the results to be less believable. Example: HD 6870's and HD 6850's test faster than a HD 5870 even though in most games this is not the case. The reason is the HD 6870 and 6850 are clocked higher than the HD 5870 and some of the 3d tests like higher clock rates over more Stream processors, Texture Units and ROP's. BTW I was going to use Badbonji's example aka 5770/5870 but the results are not showing the 5770 as being faster.

So what does this all mean. Passmark PerformanceTest is a synthetic test. No synthetic test can be 100% accurate when compared to real world testing.The only good way to test graphics cards is by testing them in actual games or programs you are going to use. That is where good hardware review sites come in. Is PerformanceTest any good? IMO its a great way to test changes on a given system, OC'ing, tweaking, hardware changes etc. It can also be used to compare two or more systems to each other in general terms though some knowledge and interpretation may be needed. PerformanceTest 8 is in internal beta testing and should hopefully have updated testing in several areas. GPU, CPU and storage tests should be greatly improved to better support new tech. Video playback testing may also be included but this was only hinted at. It might not be possible do to the constant changes being made to file formats, codec's etc.

Oh, one of my favorite features in Performancetest is comparing my past systems to see the difference in performance over the years. Performancetest has kept compatibility between versions over the years though some tests have been added so some old tests done with older versions have no results for some tests. Also some test results seem strange. Example the 2d results are huge for the 9700 pro. This may very well be accurate as newer cards seem to have forgotten about 2d performance. OS may also play a part in some results.   See the png below.

Bill

[Dark Mantis]

Hi Bill

Thanks for posting the charts, very interesting especially as you pointed out the 2D decline. I wish I had kept all my old records of old systems but they seem to have been lost along the way.

[Badbonji]

My bad, I now see it was the mobility version! I think that benchmark isn't big enough to be supported, 3dMark's benchmarks are more popular for comparing hardware. The 3d benchmarks apart from the highest quality is very sensitive on the CPU used as well, although clock speed seems to be more important than cores in your screenshot. The rest of the system will also be important in determining the score.

As for my degree, it is only a single one (The majority are in the UK). I am studying towards an MSci in Physics at the University of Birmingham. Yeah I am very glad that I chose to go to uni, I don't think I will have as much free time ever again afterwards!