Best Graphics Cards For The Money: June 2010

[cleeve]

True, but then such extra costs are not really that much these days given so many boards do support CF and/or SLI at good prices (especially for older CPUs such as the Q6600, eg. Asus P5N 32-SLI Premium is about 35 UKP), and gfx cards now use so much power that a decent PSU is a wise buy anyway.

Ah, but we're not comparing today's prices. Your argument assumes we're comparing prices from two years ago.

Either way, like you say, I've obtained a good speedup for less than 50 UKP, reaching a level that will beat a 4890 or 5770 in many cases.

This I'm unclear on. Does this mean you've had the first 8800 GT for two years, and only recently bought a second card for SLI? If this is the case, nobody will criticize that move, if you've already got an 8800 GT and an SLI mobo, then buying a second 8800 GT is the best bang for your buck upgrade you can do. But If you're starting out right now--based on on US pricing, and I don't have the time to be a UK pricing expert--I'd choose dual 4850s or GTS 250 cards. Far better value there.

... Why are you assuming *your* prices are the only ones that matter?

They're the only ones that matter to me. I never said they were the only ones that mattered in some objective sense.

Fair enough, but if this is the case don't criticize me for quoting the prices that matter to me.
Jabs like "the world is bigger than the USA" aren't necessary, I'm sure you could think of a nicer way to make the point.

Anyway, my new question. Am I right in thinking that a Radeon 4890 should stomp on an 8800GT? (single cards I mean, never mind 8800GT SLI) I thought this
would be the case, but my friend's system - an i7 930 with one (now two) 4890s - isn't seeing the kind of performance I would expect. See:

http://www.sgidepot.co.uk/misc/pctests.html

I know these are 3DMark06 results and so the overall scores are not that meaningful since they're skewed by CPU speed, but the fps scores for the individual tests (Proxycon, Firefly, Canyon, Freeze) ought to be at least vaguely useful comparisons, yes?

Depends on your objective. 3DMark is almost completely useless for people making purchasing decisions, it is inconsistent and misleading. At best it's a vague relative performance gauge but frankly I put no stock in 3D Mark results, it is rare that game performance correlates closely to 3DMarks. If you want to know how cards perform relatively, you need game benchmark results.

If you're using it as a tool to see if his rig is running correctly, then it might be a bit more useful yes--but only against other machines with the exact same graphics setup. But in order for the data to be useful you first need to be exactly sure that the tests you're comparing were run at the exact same settings.

I'd be more interested in seeing actual benchmark data in your friend's case. It would be a much better indicator if something was wrong or not. The Stalker: Call of Pripyat down-loadable benchmark will be more useful for this kind of troubleshooting.

 

[Guest]

The 4870X2s are gone I guess you could look at used ones but that is kind of risky. The 4870 are still in stock tough (http://www.newegg.com/Product/Product.aspx?Item=N82E16814150436&cm_re=4870-_-14-150-436-_-Product). Yes the 5970 is about has powerful as 2 5970 but it's not necessary two 4870s have about the same performance has the 5770 so it would probably equal two 5770 in Xfire which has about the same performance has the 5850.

So I could use your suggested card from newegg to go with my card here http://www.tigerdirect.com/applications/SearchTools/item-details.asp?Sku=P450-4880 with no issues?

 

[Userremoved]

So I could use your suggested card from newegg to go with my card here http://www.tigerdirect.com/applications/SearchTools/item-details.asp?Sku=P450-4880 with no issues?

It should. It's the same GPU.

 

[mapesdhs]

Cleeve writes:
> This I'm unclear on. Does this mean you've had the first 8800 GT for two years, and only recently bought a second card for SLI? ...

For my main gaming setup, yes. I then bought two eBay ASUS 8800GTs for my new i7 860 setup; these are better/better/worse/better respectively than
my friend's 2x4890 setup for the individual 3DMark06 tests (Proxycon/Firefly/Canyon/Freeze), thus my confusion. Like you say though, I'll try the COP
test instead, perhaps that will give more logical results.


> ... based on on US pricing, and I don't have the time to be a UK pricing expert--I'd choose dual 4850s or GTS 250 cards. Far better value there.

Maybe it's just pricing variance then. Atm two 4850s or GTS 250s would cost at about 200 UKP here (260 for good ones), though I notice most places
don't have 4850s in stock. There's a lot of overlap with 5770 pricing though, so perhaps two 5770s would be better (260 is enough for two XFX cards).
If one has that budget of course.


> Jabs like "the world is bigger than the USA" aren't necessary, I'm sure you could think of a nicer way to make the point.

Yes, you're right. Sorry about that!



> relatively, you need game benchmark results.

I don't know what games he has, but I suggested trying some of the downloadable game demo benchmarks, eg. Stalker COP.


> I'd be more interested in seeing actual benchmark data in your friend's case. It would be a much better indicator if something was wrong or not. The
> Stalker: Call of Pripyat down-loadable benchmark will be more useful for this kind of troubleshooting.

Ok! I'll pass that on. I ran it on my system early on, but that was before sorting out the CPU oc and SLI stuff.

Hmm, perhaps best if I wait until he's sorted his own CPU oc (he also has an U120E/SFF21F). He should get a decent oc since he's starting out with a 930.

Thanks!!

Ian.

 

[Userremoved]

Cleeve writes:
> This I'm unclear on. Does this mean you've had the first 8800 GT for two years, and only recently bought a second card for SLI? ...

For my main gaming setup, yes. I then bought two eBay ASUS 8800GTs for my new i7 860 setup; these are better/better/worse/better respectively than
my friend's 2x4890 setup for the individual 3DMark06 tests (Proxycon/Firefly/Canyon/Freeze), thus my confusion. Like you say though, I'll try the COP
test instead, perhaps that will give more logical results.


> ... based on on US pricing, and I don't have the time to be a UK pricing expert--I'd choose dual 4850s or GTS 250 cards. Far better value there.

Maybe it's just pricing variance then. Atm two 4850s or GTS 250s would cost at about 200 UKP here (260 for good ones), though I notice most places
don't have 4850s in stock. There's a lot of overlap with 5770 pricing though, so perhaps two 5770s would be better (260 is enough for two XFX cards).
If one has that budget of course.


> Jabs like "the world is bigger than the USA" aren't necessary, I'm sure you could think of a nicer way to make the point.

Yes, you're right. Sorry about that!



> relatively, you need game benchmark results.

I don't know what games he has, but I suggested trying some of the downloadable game demo benchmarks, eg. Stalker COP.


> I'd be more interested in seeing actual benchmark data in your friend's case. It would be a much better indicator if something was wrong or not. The
> Stalker: Call of Pripyat down-loadable benchmark will be more useful for this kind of troubleshooting.

Ok! I'll pass that on. I ran it on my system early on, but that was before sorting out the CPU oc and SLI stuff.

Hmm, perhaps best if I wait until he's sorted his own CPU oc (he also has an U120E/SFF21F). He should get a decent oc since he's starting out with a 930.

Thanks!!

Ian.

You do know that the 4890 is more powerful than the GTX260 which WAY better than 8800Gt. Look at the benchmarks not at 3D mark 06.

 

[Userremoved]

You might want to look at this look at how much better this card is compared to the 9800Gt (which just a re branded 8800GT).

 

[mapesdhs]

Userremoved writes:
> You do know that the 4890 is more powerful than the GTX260 which WAY better than 8800Gt. Look at the benchmarks not at 3D mark 06.

Know? This is why I've been asking questions, as my friend's test results don't make sense. Did you even look at my page before
posting that reply? See: http://www.sgidepot.co.uk/misc/pctests.html

I've already read lots of reviews of the 4890 btw. Indeed, I've been nagging my brother for months saying he should get one, though
it looks like it's too late now.


Note that I'm not referring to the overall '06 scores, those are obviously skewed by main CPU speed and are thus meaningless.
Look at the individual 'game' scores instead, eg. Test 4 vs. Test 12, ie. the fps numbers for Proxycon/Firefly/Canyon/Freeze.

However, Cleeve is probably right, there's likely something just bogus about these 3DMark tests in how they work (I hope), so with
any luck the Stalker test will be more revealing. Meanwhile though, plenty of people do use '06 results as part of their decision
making, so if the results I'm seeing are actually 'normal' for how '06 works then it's still good to know this is the case.

Ian.

PS. Can you trim your quoting please? There's no need to include my entire original post.

 

[Userremoved]

^Ok then sorry would is second PCIe happen to be at 4X?

 

[mapesdhs]

You might want to look at this look at how much better this card is compared to the 9800Gt (which just a re branded 8800GT).

I assume you know that fps scores that high don't mean much either, especially at such a low res and low detail. That graph conveys nothing useful to me at all.

Ian.

 

[mapesdhs]

Btw, here's a good article:

http://www.hardwarecanucks.com/forum/hardware-canucks-reviews/33190-gpu-benchmarking-methods-investigated-fact-vs-fiction.html

Ian.

 

[Userremoved]

So did you find why the 4890s did not work at full power?

 

[gm0n3y]

Why are isn't the 4850 included in the latest Tom's charts?

 

[mapesdhs]

So did you find why the 4890s did not work at full power?

Not yet; I'm waiting for my friend to run the Stalker COP tests, but first he wants to sort out the oc of the 930 chip; he's using an
Asus P6X58D Premium mbd with a Fenrir cooler, 6GB RAM (not sure of the speed), Thermaltake Toughpower 750W PSU.

Meanwhile, I'm going to redo some of my tests so the CPU clock is always the same, make them more comparable.


Did you read the article on hardwarecancucks btw? It's very interesting.

Ian.

 

[lunosnyper]

[citation][nom]The_King[/nom]You running three 3870s ? I think you maybe got a 3870x2 and and 3870 x-fire setup I don't think you going to get much of an improvement with the $200 budget since the 3870x2 has the similar performance to the 5830 which is $220.If your budget excludes selling the 3870's try selling them and going for a 5850[/citation]

A 3870x2 is about the same as a 1gig 4850 i don't know what benchmark your looking at. So i would assume that 4850's CF would kick 3870x3's ass.

 

[Diabolical User]

[citation][nom]pirateboy[/nom]how about writing some real articles instead of this "best something for the money" B.S.[/citation]

You must be the kind of person that lives for advertisements and thrives off of being swindled out of your money; all for the purpose of having a "shinier" sticker.

 

[Guest]

I got a 4850 recently and i can play all the games on the highest details available.... what more do we need? will buy the 5970 when it touches 200$ after 2 yrs i guess hehe ....

 

[Userremoved]

[citation][nom]bharatwd[/nom]I got a 4850 recently and i can play all the games on the highest details available.... what more do we need? will buy the 5970 when it touches 200$ after 2 yrs i guess hehe ....[/citation]
But then there will be the new HD 8970.

 

[Guest]

wait a minute , i think 2x hd5750 means the universal shader is 720x 2, slightly lower than 2x hd5770 which is 700x2. the table on page 4 s incorrect.

 

[cleeve]

[citation][nom]yoga[/nom]wait a minute , i think 2x hd5750 means the universal shader is 720x 2, slightly lower than 2x hd5770 which is 800x2. the table on page 4 s incorrect.[/citation]

Good catch there, fixed!

 

[logainofhades]

There does not seem to be any real reason to upgrade my 4870 1gb. The price for the performance increase just does not make it worthwhile.

 

[prljamm1]

Which of these would work well on a Dell Dimension E521, playing the "Total War" games? Thanks.

 

[silicondoc]

This whole thing is a JOKE. There isn't a single 5750 anywhere for $125. I noticed every single ati card is listed at alower price than is actually available. I'm sure there are the single rare exceptions.
On the other hand, the nvidia cards are priced higher than they actually are in the article.
Worse yet you go to the high graphics gaming charts and not a single GTX400 series is listed to this very day.
This site is a biased joke now.
What a shame.

 

[cleeve]

[citation][nom]silicondoc[/nom]This whole thing is a JOKE. There isn't a single 5750 anywhere for $125. [/citation]

$124.99 ($105 after mail in rebate!)

http://www.newegg.com/Product/Product.aspx?Item=N82E16814131331&cm_re=5750-_-14-131-331-_-Product

Whose the joke, again?

 

[Guest]

Not sure if anyone is even paying attention to comments on this article, but I'm curious as to why the GTS250 is mentioned in the $100 category but no mention of dual GTS250s. There's a crazy sale going on at Newegg right now that puts this card at around $60 bucks if you include rebates and combo deals...