Best Graphics Cards For The Money: June 2010

Page 4 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.
G

Guest

Guest
As some users point out you cand buy a very cheap second hand board whit great price paerformance value i myself have buy a gt240 for 70$ in juanary this year and i nver looked back if i overclok it is clearly above the 9600GT whit low power and beter performance :d
 

mapesdhs

Distinguished
Hate to say it but I've still yet to see any product that offers
a significant speed increase worth the expense compared to using
two 8800GTs in SLI, except for those games where the 512MB RAM
limit in these cards is a factor.

One can get 8800GTs 2nd-hand for about $60, or perhaps $80 for
the more popular brands. For my new i7 build, I bought an ASUS
Glaciator for 46 UKP and then another for 49 UKP (I really like
these cards, they make very little noise), though I may use
two Gigabyte Turboforce cards instead which use faster stock
clocks (I already had one Gigabyte, bought another for 40 UKP).
Overall, very little cost and the speed is impressive. With
initial testing, I get 22488 3DMark06 for the two ASUS boards,
which is significantly faster than a 4890 (I should get even
better results for the two Gigabyte boards). See:

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

Indeed, it even beats 4890 CF for many tests, though I do wonder
about my friend's system which has the two 4890s in CF, I was
expecting the scores to be better for that config, though the
Feature tests appear consistent, eg. multi fill rate went up from
22622 to 53864.

And btw, for those who claim the P55 platform is bad for SLI/CF,
that's just plain wrong. I'm using an Asrock P55 Deluxe and the
SLI setup is working very well indeed, I'm extremely pleased. I
wanted a gfx setup that could play older games (Oblivion, Stalker
SHOC, CoD4, etc. at high-res max-detail, ie. either 2048 x 1536
or 2300 x 1440 (the latter is a 24" wide CRT, Sony GDM-FW900).

Note my initial query: what single newer card is worth the
expense at the $200 pricepoint? The 5850 would cost 2X more
than two 8800GTs, yet would be slower from the results I've
read so far.

DX11 support, etc. is a different issue. I'm only talking about
performance here. IMO, the current crop of gfx cards are just
too expensive. When I bought my first 8800GT for 120 UKP (12/May/2008),
it gave a huge speed increase over my old X1950Pro. More than two
years later, what card is there available now at this price point
which is significantly better? There isn't one IMO.

Unless one is playing games that need the extra RAM, either by
default or because one is using a high-res display, or one wants
modern features like DX11/Eyefinity, there's a heck of a lot of
value in buying older cards like the 8800GT and using them in SLI/CF.

Ian.

 

katokato

Distinguished
May 1, 2010
47
0
18,530
Yay, glad to see the 5770 on the list, i have just bought one.

I cant believe that the 4850 is still there, keep going ATI!
 

cleeve

Illustrious
[citation][nom]mapesdhs[/nom]Hate to say it but I've still yet to see any product that offersa significant speed increase worth the expense compared to usingtwo 8800GTs in SLI[/citation]

Two 4850's in CrossFire fits that bill nicely.

[citation][nom]mapesdhs[/nom]One can get 8800GTs 2nd-hand for about $60, or perhaps $80 for the more popular brands.[/citation]

And you can get a brand new 4850--with a warranty--for $100.

Nevertheless, used is a whole different ballpark. This list is obviously for new cards, and used card are cheaper but that comes at it's own price: lack of warranty in most cases, and lack of knowledge as to how hard the previous owner pushed the card before dumping it.

Nobody here said the used market isn't viable, but it's not the focus of this article. It's not feasible to write an article for the used market because it doesn't exactly follow any rules. If someone wants to sell you a used 4870 for $20, there's nothing stopping them... but just because someone sells one at that price doesn't mean you can expect anyone else to, either.


 

mapesdhs

Distinguished
Cleeve writes:
> Two 4850's in CrossFire fits that bill nicely.

Alas, new pricing for that is too high.


> And you can get a brand new 4850--with a warranty--for $100.

(more like the equivalent of $130 here)

Yes, but that doesn't get me the performance level I want. I'd
need two, which is $260, more than I want to spend. My point
is that 2 years ago one could buy a gfx card at the $150 price
point which *did* offer a huge speed boost over the equivalent
card from a year or 2 before that. Today this is not true at all.
After 2 whole years, the 4850 offers no useful speed increase
over an 8800GT, and (for example) site reviews show 4850 CF is
slower than 8800GT SLI for Crysis, etc.

Looks like the first useful card in terms of extra speed over
an 8800GT is a GTX260, but that's more like $240 here.

Other viable cards are no longer available, or cost too much.


> Nevertheless, used is a whole different ballpark. This list is
> obviously for new cards, and used card are cheaper but that
> comes at it's own price:

Absolutely true. I guess my point is a double one: used cards
offer excellent performance, while at the same time, at the
price point I mentioned, new cards today offer nothing useful
compared to what was on offer two years ago for those looking to
upgrade at that time. It's as if the whole pricing bracket has
moved upwards. I can remember not that long ago review sites
crying merry hoohaa over the idea of a single gfx card costing
$700+, but today nobody minds at all.

And don't even get me started on power consumption. ;D


> ... and lack of knowledge as to how hard the previous owner
> pushed the card before dumping it.

Very true, it's a risk. I'm careful when I look at what's
available. The only thing that really surprises me about 2nd
hand is the number of people who seem happy to buy cards that
are described as faulty... bizarre!


> ... It's not feasible to write an article for the used market
> because it doesn't exactly follow any rules. ...

Yes, I agree. What I really mean is, why is it now the case
that $150 doesn't get one the kind of speed increase over a
2-year-old card compared to what was available for this amount
of money 2 years ago? IMO the pricing's gone whacko.

Naturally my strategy would be a tad different if I was looking
to play newer games which do need a lot more RAM, but that isn't
the case yet. It's just wierd that here I am (or was) with around
100 UKP spend on a new gfx card and there's just nothing available
which is any good. 2 years ago that price got me an 8800GT and
a 100% speed increase over my previous X1950Pro, a card which
when it was launched was widely praised. (and btw, my AGP version
gave better 3DMark06 scores than review sites were getting with
the PCIe version).

CPU speed has increased nicely in the last 2 years, but GPU
speed hasn't at all (at equivalent price points I mean).
Afterall, as review sites themselves said often enough, how many
times has the 8800GT GPU been rebadged & remarketed? 9800, then
the 250, etc.

So what's changed in the marketing of gfx cards in the last 2
years? Is everyone richer and vendors are just aiming higher as
a result? I doubt it.

Ian.

 

cleeve

Illustrious
[citation][nom]mapesdhs[/nom](more like the equivalent of $130 here)[/citation]

Well, geography certainly does have an impact. But in the US you can get two 4850's for $200 online brand new, so your argument falls apart here as two 4850's will eat two 8800 GTs alive.

Of course the world is a big place, pricing--and therefore value--certainly varies from country to country.
 

mapesdhs

Distinguished
Cleeve writes:
> ... as two 4850's will eat two 8800 GTs alive.

Nope, that's completely wrong according to reviews and numerous
forum posts (where did you get that idea from? :D) eg. see:

http://www.anandtech.com/show/2553/14
http://www.xtremesystems.org/forums/showthread.php?t=191944

etc. Lots of examples around, though alas tom's 4850 didn't
include 8800GT numbers.

In the Anand tests though, the two configs are almost identical
for Crysis and Assassin's Creed; the 4850 CF setup wins for COD4
and Oblivion (latter with a small margin); 8800GT SLI setup wins
for Enemy Territory: Quake Wars.

In all cases however, the difference is not that great. At best
it's about 30% (COD4), but for Crysis there's almost no difference
(8800GT SLI slightly quicker).

So no 'eating' going on I'm afraid. Do people believe 4850 must
be quicker because they want to believe it's quicker? Blimey,
Goodkind was right... ;)

Anyway, that's more relating to my comments about 2nd-hand items,
etc. My bigger point is the general business of pricing as a
whole, ie. we're expected to spend a lot more now to get a
product which gives a 2X speed boost over whatever was current
2 years earlier. Why?

Ian.

 
G

Guest

Guest
I'm wondering about the price shifting going on, everyone is talking generation to generation improvement and the cost to upgrade, but I bought a 4850X2 for $235 from the egg and I'm wondering when and why did prices double for equivalent performers.
 

cleeve

Illustrious
[citation][nom]mapesdhs[/nom]Nope, that's completely wrong according to reviews and numerousforum posts (where did you get that idea from? ) eg. see: http://www.anandtech.com/show/2553/14 [/citation]

Nope. Completely wrong yourself, sport. You have to look at the information in context. :)

3d marks don't count for shizz, and the Anand review is using launch 4850 drivers vs. mature 8800 drivers in that 4850 launch review. The catalysts--and CrossFire--have come a very long way since then, and even in the beginning the 4850's were meeting or beating the 8800s in SLI...
 

mapesdhs

Distinguished
Cleeve,

I wasn't looking at 3DMark numbers at all. I was checking real
game fps scores. And it's not just Anand's results, I checked
a lot of sites & posts.

Besides, that doesn't answer my main point about pricing, which
is more important.


vogonpoetry,

Pretty much anything. :D

Ian.

 
G

Guest

Guest
what if my monitor only goes up to 1024x768? What card will get me good performance then?
 

cleeve

Illustrious


Odd. That Xtremesystems link you provided seems to showcase 3DMark scores more than anything else.




You keep mentioning '2 years'.

Are you aware that in June 2008, GeForce 8800 GTs were $170 a pop? That's more than 2 used 8800 GTs today! :p

Now, $160 or so gets you a Radeon HD 5770 that's a lot faster. It seems to me that your main point about pricing doesn't quite pan out when you look at the facts... ;)
 

mapesdhs

Distinguished
Cleeve writes:
> That Xtremesystems link you provided seems to showcase 3DMark scores more than anything else.

There are plenty of other pages/examples. Google is your friend. :)


> You keep mentioning '2 years'.

It's a typical time frame for Moore's Law to be worth examining.


> Are you aware that in June 2008, GeForce 8800 GTs were $170 a pop? ...

Give or take; mine was somewhat less than that.


> That's more than 2 used 8800 GTs today! :p

Nonsense. My new 8800GT cost 120 UKP. Two used 8800GTs cost me about 90, and I wasn't even
being particularly patient. Yesterday an ASUS Glaciator sold for 37 (really nice card). Or if one is
content with brands such as POV, Palait, Zotac, etc. then they're often as little as 35 each.

Anyway, you seem to be obsessing about my new vs. used comparison, rather than the more
important point about pricing levels.


> Now, $160 or so gets you a Radeon HD 5770 that's a lot faster. ...

Here it's more like $190.

Yes, it's faster, but enough faster if one already has something like an 8800GT? Nope. Better/cheaper
than two used 8800GT SLI? No.

My point is, the upgrade outlay only gets one about a 30% to a 50% speed increase. 2 years ago the
same upgrade outlay got one a 100% speed increase. You have to spend _more_ now to get the same
kind of speed boost. The price brackets have moved upwards. Why does nobody seem to care? The
product lines are shifting as if we're all richer, but in the middle of a global recession that seems a bit
wierd to me.


> It seems to me that your main point about pricing doesn't quite pan out when you look at the facts... ;)

The world is bigger than just the USA. :D Have a look at scan.co.uk, a typical supplier. 5770s are not $160 here.

It's a good card, yes, but not good enough.

Ian.

 

cleeve

Illustrious


Sure is, I only mentioned it because it represents half of the links you offered to support your argument. :D



You do realize that moore's law isn't really a law, right? It's more of a quip that the media has latched on to. Nobody has legislated a 200% performance increase in 2 years as far as I'm aware.




Personally, I see a massive difference in performance comparing an 8800 GT to a 5770, especially at higher resolutions with AA enabled. And I have to note that a couple 8800 GTs in SLI example adds a price premium for a larger PSU and the SLI feature on a motherboard, costs that you aren't factoring in.

I guess you feel that nobody seems to care because we're seeing value in these new cards and the added performance that comes along with them. If you don't want to upgrade your 8800 GTs, that's certainly your prerogative and I'm happy you're pleased with your setup. They remain a powerful option and it's great that you're pleased with your purchase.. But that doesn't mean that other folks don't see the value you're missing in new products.




The world is indeed bigger than the USA... hell, I don't even live there. But the world is bigger than the UK, too, so why aren't you investigating worldwide prices and submitting an aggregate to support your argument? Why are you assuming *your* prices are the only ones that matter?

Since we're commenting in an article's thread that is based on US pricing exclusively, I think US pricing is probably reasonable to consider in this context. ;)


 
G

Guest

Guest
Hey I have been researching new graphics cards for my new build. I am going with Phenom II 965 based system, and I have bought a XFX Radeon HD 4870 XXX 1GB PCIe w/Dual Link DVI for my system now and going by your chart only an HD 5970 would be a good uppgrade or maybe the 4870 x2 but I cant find it anywhere.

Would you recommend anything else in a crossfire setup that would be cheaper than the 5970 and would out perform my current card?

Also, my current card is not available either or I would order another and crossfire them.

Any help is appreciated.
 

cleeve

Illustrious
Would you recommend anything else in a crossfire setup that would be cheaper than the 5970 and would out perform my current card?

Also, my current card is not available either or I would order another and crossfire them.

CrossFire is quite forgiving when it comes to different types of cards. Look on ebay and get any 1GB 4870, or get any new 1GB 4890. Either should crossfire fine with your current card, but if you're buying used I'd look for the identical model anyway. Just make sure your motherboard and power supply supports CrossFire.

Other than that, there's no real good option for you to upgrade to that isn't a waste of $$ for the difference you'll see.
 

mapesdhs

Distinguished
Cleeve writes:
> You do realize that moore's law isn't really a law, right? ...

Of course. It's just a useful/interesting comparison of technological trends.


> ... It's more of a quip that the media has latched on to. Nobody has legislated a
> 200% performance increase in 2 years as far as I'm aware.

(NB: A 2X fps increase is a 100% performance increase, not 200%)



> Personally, I see a massive difference in performance comparing an 8800 GT to a 5770, especially at higher resolutions with AA enabled.

Must confess I'd been focusing more on NVIDIA cards in my research as I'd not been happy with ATI drivers in the past. However, the
5770 does look good, yes. How would you say a 5770 compares to a 4890 though? Still looks like 8800GT SLI will match it in many cases,
based on reviews I've read where the 5770 is often slower than a 4890.

This leads me to a different question which perhaps you can help with (see below).


> And I have to note that a couple 8800 GTs in SLI example adds a price premium for a larger PSU and the SLI feature on a motherboard,
> costs that you aren't factoring in.

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.


> ... your 8800 GTs, that's certainly your prerogative and I'm happy you're pleased with your setup. They remain a powerful option and it's
> great that you're pleased with your purchase.. But that doesn't mean that other folks don't see the value you're missing in new products.

I entirely agree, I mentioned some caveats before. In my case it was partly budget, partly because of the games I want to play atm, etc. 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.


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

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


> Since we're commenting in an article's thread that is based on US pricing exclusively, I think US pricing is probably reasonable to consider in this context. ;)

These products are available globally, and people comment & post from everywhere. It's perfectly reasonable to ask things from my own perspective based
on where I live.


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?

I did ask my friend to check his results were done with 'performance' rather than 'quality' settings; they weren't, but he's redone the tests and the new data is
not much better. I'm sure something must be wrong with his setup, but not sure what to suggest. Any ideas? Meanwhile, I found this thread:

http://forums.techpowerup.com/showthread.php?t=93531

My friend's 4890 scores definitely don't look right, especially for 2x4890 CF, eg. compared to his i3/3850 results, the 4890s don't seem to be scaling properly
for Proxycon, Firefly and Deep Freeze. Anyone have any suggestions I can pass on?

Ian.

 

mapesdhs

Distinguished
Ah, it's the old let's be rude and insulting instead of just answering my questions, etc. I expected better here.

Cleeve doesn't agree with the things I've said, but at least he's always polite.

Perhaps it's just too much of an intillectual effort for you to actually examine my friend's 4890 data and try and work out
why the results are lower than one would expect them to be, based
on Cleeve's comments and your own obvious disdain for the 8800GT.
I can only infer this is the case if you're able to do nothing except insult me.

The query still stands: my friend's system should stomp on mine,
from what people have said here, but it doesn't. Why? Is it the
nature of the tests? If so, then in what way? If not the tests,
then what advice might I pass on to help his 4890 perform as it
should?

Perfectly reasonable questions IMO.

Ian.

 

Userremoved

Distinguished
Feb 27, 2010
1,276
0
19,310
[citation][nom]TonyGriff[/nom]Hey I have been researching new graphics cards for my new build. I am going with Phenom II 965 based system, and I have bought a XFX Radeon HD 4870 XXX 1GB PCIe w/Dual Link DVI for my system now and going by your chart only an HD 5970 would be a good uppgrade or maybe the 4870 x2 but I cant find it anywhere. Would you recommend anything else in a crossfire setup that would be cheaper than the 5970 and would out perform my current card?Also, my current card is not available either or I would order another and crossfire them.Any help is appreciated.[/citation]
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.
 

Userremoved

Distinguished
Feb 27, 2010
1,276
0
19,310
[citation][nom]Userremoved[/nom]I'm sorry but the 8800GT is getting really old. It was great in 07 but now...Not so much. The 4850s still are string cards and considering the 4850 performs like the GTS250 which a re branded 9800GTX which is a re branded 8800GTX the two 4850s will perform slightly better and in some games the 8800GT performed better because there coded for the nVidia cards.[/citation]
*Strong
 

Userremoved

Distinguished
Feb 27, 2010
1,276
0
19,310
[citation][nom]mapesdhs[/nom]Cleeve writes:> You do realize that moore's law isn't really a law, right? ...Of course. It's just a useful/interesting comparison of technological trends.> ... It's more of a quip that the media has latched on to. Nobody has legislated a> 200% performance increase in 2 years as far as I'm aware. (NB: A 2X fps increase is a 100% performance increase, not 200%)> Personally, I see a massive difference in performance comparing an 8800 GT to a 5770, especially at higher resolutions with AA enabled.Must confess I'd been focusing more on NVIDIA cards in my research as I'd not been happy with ATI drivers in the past. However, the5770 does look good, yes. How would you say a 5770 compares to a 4890 though? Still looks like 8800GT SLI will match it in many cases,based on reviews I've read where the 5770 is often slower than a 4890.This leads me to a different question which perhaps you can help with (see below).> And I have to note that a couple 8800 GTs in SLI example adds a price premium for a larger PSU and the SLI feature on a motherboard,> costs that you aren't factoring in.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 (especiallyfor 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 PSUis a wise buy anyway.> ... your 8800 GTs, that's certainly your prerogative and I'm happy you're pleased with your setup. They remain a powerful option and it's> great that you're pleased with your purchase.. But that doesn't mean that other folks don't see the value you're missing in new products.I entirely agree, I mentioned some caveats before. In my case it was partly budget, partly because of the games I want to play atm, etc. Eitherway, 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.> ... 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.> Since we're commenting in an article's thread that is based on US pricing exclusively, I think US pricing is probably reasonable to consider in this context. These products are available globally, and people comment & post from everywhere. It's perfectly reasonable to ask things from my own perspective basedon where I live.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 thiswould 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.htmlI 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 theindividual tests (Proxycon, Firefly, Canyon, Freeze) ought to be at least vaguely useful comparisons, yes?I did ask my friend to check his results were done with 'performance' rather than 'quality' settings; they weren't, but he's redone the tests and the new data isnot much better. I'm sure something must be wrong with his setup, but not sure what to suggest. Any ideas? Meanwhile, I found this thread: http://forums.techpowerup.com/showthread.php?t=93531My friend's 4890 scores definitely don't look right, especially for 2x4890 CF, eg. compared to his i3/3850 results, the 4890s don't seem to be scaling properlyfor Proxycon, Firefly and Deep Freeze. Anyone have any suggestions I can pass on?Ian.[/citation]
I'm sorry but the 8800GT is getting really old. It was great in 07 but now...Not so much. The 4850s still are string cards and considering the 4850 performs like the GTS250 which a re branded 9800GTX which is a re branded 8800GTX the two 4850s will perform slightly better and in some games the 8800GT performed better because there coded for the nVidia cards.
 
Status
Not open for further replies.