Originally posted by Quizzical

Originally posted by Mako89

If I would be you I would go for AMD HD 7770 Crossifre. Why? Becouse a pair of 7770 in crossfire scale verry well + both of em are gonna cost you like 300$.

One higher end card is nearly always better than two lower end cards in CrossFire or SLI, as the single card lets you skip all of the problems that CrossFire and SLI bring.  Average frame rates don't tell the whole story.  CrossFire and SLI really only make sense for people who think one high end card isn't good enough, and so they'll get two high end cards.  The original poster sure doesn't seem like the sort of guy who insists that a single $550 Radeon HD 7970 doesn't give enough performance, and is looking to spend more on something else instead.

Speaking from experience, this post, right here^^^^^^ is TRUE. DO NOT, i repeat DO NOOOOT go with crossfire if you have other options. I can't stress it enough.

Only Nvidia:

if you can spend only  less than 200$ go for 550 Ti

if you can spend 200-300$ go for  gtx560Ti-gtx 570 (570 highly recomended)

if you want to spend over 400$ go for gtx580 or580 SC (Best upgrade you can do for the price)

Inever use Ati so thas why i dint post any of theyr graphic card  tho im not saying is not good or somthing mi recomendation are Nvidia-EVGA-MSI.

peace

Originally posted by dadante666

Only Nvidia:

if you can spend only  less than 200$ go for 550 Ti

if you can spend 200-300$ go for  gtx560Ti-gtx 570 (570 highly recomended)

if you want to spend over 400$ go for gtx580 or580 SC (Best upgrade you can do for the price)

Inever use Ati so thas why i dint post any of theyr graphic card  tho im not saying is not good or somthing mi recomendation are Nvidia-EVGA-MSI.

peace

Your recommendation seems to be basically to ignore prices.  A GeForce GTX 560 Ti or GeForce GTX 570 can sometimes be a reasonably good value for the money.  A GeForce GTX 550 Ti or GeForce GTX 580 pretty much never are.

Originally posted by Ridelynn

 

---

Originally posted by fiontar

---

Originally posted by Quizzical
That should be more than double the performance of a GeForce GTX 260.  It's a reasonably good value if you regard rebates as "free" and don't care much about power consumption or newer features, but not so much otherwise.

---

It's a non-reference design with better performance than a 560Ti with lower power consumption, for $20 more. Since you've been recommending the 560Ti to some people, not sure why you would see this as a lesser value?

---

 

No, a 570 is not lower power consumption than a 560Ti.

If you go with the argument that it has similar performance to a 7870, then yeah, it seems like your saving about $70, as the 7870 still costs about the same as the original MSRP on the 570 did.

But there are other people that think a 7850 is also competitive, especially with overclocking. And it's considerably lower power, and for just a bit less (~$260 with no rebates).

The 560Ti at $200 is a good deal. The 570 at $280 is not a bad deal, it's just that the 7850 is nipping on it's heals, but compare stocks and the 570 is the faster card, and isn't that much more expensive (under $300). The 7870 at $360 is... not really a good deal, AMD is price gouging on that one right now, but they can afford to - because it offer a lot besides just performance over the 570.

But the 570's aren't going to stay that price forever - stock will run out soon (as I think that's what is happening, we are seeing a huge inventory reduction before the 670 introduction), and the price on the 7870's won't stay that high forever either, as once nVidia finally rolls out a competitive card, they will have to adjust prices.

I can see going either way. I think that the current <$300 prices on the 570's are a great deal (kind of like when we had the great 5850 prices a couple of summers ago), but I think they are very temporary, and that the 670 will be out pretty soon, and when that hits, I expect some price shuffling with the 7870 and 7950 to compete. I know personally, I would be happy that I got a great deal on a good video card, and somewhat bummed that 2 weeks later the upgraded version came out for not really that much more money...

I understand weighing the short discount pricing window on the previous generation cards vs. the anticipation of soon to be released next-gen cards. It becomes a bit of a crap shoot. I already stretched my budget slightly to get this GTX 570. I figured that the GTX 670 is likely to be priced similarly to the GTX 570 suggested retail price at it's launch. Add in the premium for scarce cards during the first couple months of availability and I'm assuming the GTX 670 will be well out of my price range in the coming months. If the next gen card in my price range is a new GTX 660, even though it will certainly outperform a GTX 560, the performance advantage over the GTX 570 is not guaranteed.

I do avail myself of rebates, so I do figure the rebate into my final card cost. With the discount window likely to be fairly short as old stock is depleted, I'm thinking I may not take much of a loss IF I should decide to turn around and sell it for a next gen card.

However, I'm expecting I will probably stick with the GTX 570 for at least a year, so the advantage of a next gen card would have to be pretty major on the price/benefit comparison to bother.

I looked at this more theoretically in February, but April was the month I planned to make a buying decision and I feel comfortable with this purchase at this price, at this snapshot point of time. :)

As to power consumption...

This non-reference design seems to be a bit more efficient than other GTX 570 cards. (Sorry for not specifying the comparison to the 560 Ti 448 core. Damn NVidia for the model number confusion. I considered buying a 560 Ti 448, not a plain 560 Ti). Also note that this card is factory clocked at 800Mhz and still manages to test at lower power consumption under load than a default clocked reference GTX 570 card. Tipping the balance as well was that this was the quad display version, but was not priced to reflect the normal premium for this model.

http://www.guru3d.com/article/kfa2-geforce-gtx-570-mdt-x4-review/7

We'll be calculating the GPU power consumption here, not the total PC power consumption.

Measured power consumption

1. System in IDLE = 162W
2. System Wattage with GPU in FULL Stress = 317W
3. Difference (GPU load) = 155W
4. Add average IDLE wattage ~ 25W
5. Subjective obtained GPU power consumption = ~ 180 Watts

Maximum power consumption is tricky to test, because if AMD and Nvidia know that it's a power consumption test, they'll have their drivers clock the card back severely to look like they're lower power.  My view is that you need to give performance figures and power consumption from the same test in order to have a meaningful result.

I'd question their methodology as well.  If you're measuring system power consumption, then that includes other components, not just the video card.  Games tend to push a processor in addition to a video card, so you get some added power consumption there.  Using a Bloomfield/X58 setup that is horrible on power consumption will make that effect a lot worse than it needs to be.  Even if they somehow found an application that doesn't touch the processor at all, you're still adding in power supply inefficiencies.  That they don't account for that (which really isn't that hard to do) is rather shocking.  So they're probably seriously overestimating how much power the video cards use in their test.

And then you should consider that that's nowhere near a real max load.  Consider, for example, that reviews of the Radeon HD 6970 found that Metro 2033 was able to make PowerTune kick in now and then.  That's with a cap of 250 W, which means that Metro 2033 was able to pull 250 W from the video card alone.  The Guru 3D test finds power consumption of 209 W, and once you subtract off the reasons why they're overestimating things, they're probably seeing under 200 W of power consumption.  That's not a max load on the card.  It might be typical power consumption for a game where the video card is the bottleneck, but the T in TDP doesn't stand for "typical".

2 GTX 460 sli! Works great and is cheap. OC fer sure

Originally posted by Quizzical

Your recommendation seems to be basically to ignore prices.  A GeForce GTX 560 Ti or GeForce GTX 570 can sometimes be a reasonably good value for the money.  A GeForce GTX 550 Ti or GeForce GTX 580 pretty much never are.

Agreed, 550 is junk and 580 is not worth the money, if you have loads of it you would get the new 680 instead.

The 570 is pretty nice for the price you can find it at if you look a little though.

Originally posted by fiontar

Hi guys. I'm looking to upgrade my GTX 260 on my current rig, looking for "best bang for the buck" in the $180 to $280 price range.

My system specs:

Win 7 Pro

AMD Phenom II X6 1055T Thuban 2.8 Ghz, OC rock solid at 4.03 Ghz

8GB Corsair XMS3 DDR1600 ram (2x4)

Corsair CMPSU-850TX 850-Watt TX Series 80 Plus Certified Power Supply

I know the price range is a bit broad, so suggestions at both ends of the range and thought on how much more performance I might expect for the extra $100 would be very much appreciated. :)

Also would be curious if any new products worth waiting for in the price range are expected in the next coule months. I could put off the upgrade for another month or two with out issue.

Thanks in advance.

This one if 5 bucks over your price range but you get a huge step for those 5 bucks. Saving a few bucks on the card will come back and haunt you later, a good card will last longer so it really cost the same in the long run and you'll get a lot better graphics for the next 2 years ( a good card last 3-4 years, a crap 2 at best): http://www.newegg.com/Product/Product.aspx?Item=N82E16814134125

Originally posted by Amjoco

2 GTX 460 sli! Works great and is cheap. OC fer sure

No it wont.

Rule number one: SLI is great if you use 2 state of the art card. A single high end card will give you more performance in the real world than 2 mid range cards. A single 570 is a lot better choice, particularly for MMO players. Many MMOs doesn't even support SLI and most of the ones that does is not doing it good enough. SLI typically is best for people that only play high end FPS games.

2 680 GTX on the other hand is fine if you can afford them. Not a choice for OP though.

Originally posted by Loke666

This one if 5 bucks over your price range but you get a huge step for those 5 bucks. Saving a few bucks on the card will come back and haunt you later, a good card will last longer so it really cost the same in the long run and you'll get a lot better graphics for the next 2 years ( a good card last 3-4 years, a crap 2 at best): http://www.newegg.com/Product/Product.aspx?Item=N82E16814134125

It's also ECS, and I wouldn't trust them for build quality.  I guess that's mainly a verdict on their motherboards, more so than their video cards, but still.  I'd sooner pay a few dollars more (and it's a difference of less than $3) for a Galaxy card, as the original poster did.

Originally posted by Loke666

Originally posted by Amjoco

2 GTX 460 sli! Works great and is cheap. OC fer sure

No it wont.

Rule number one: SLI is great if you use 2 state of the art card. A single high end card will give you more performance in the real world than 2 mid range cards. A single 570 is a lot better choice, particularly for MMO players. Many MMOs doesn't even support SLI and most of the ones that does is not doing it good enough. SLI typically is best for people that only play high end FPS games.

2 680 GTX on the other hand is fine if you can afford them. Not a choice for OP though.

Ew I wasn't comparing to anything, just said it works great and is cheap. :)

OP just a little graph. I think some others had some also posted. http://www.sweclockers.com/image/diagram/1970?k=2f29fbaae1fc8fea55a32bcd56cdda9e

Quick question for Quizzical -

I'm in a similar boat to the OP (except my card is a 240), and I was already looking at the 560ti - came across a MSI N560GTX-Ti Twin Frozr II 2GD5/OC and was wondering if the 2GB vram makes it a better choice, it's only €20 more than the 1gb version, and while more is generally better in terms of VRam, I know it's not always the case, so I thought I'd ask the resident guru.

OP - sorry for slight derail!

Originally posted by Quizzical

Maximum power consumption is tricky to test, because if AMD and Nvidia know that it's a power consumption test, they'll have their drivers clock the card back severely to look like they're lower power.  My view is that you need to give performance figures and power consumption from the same test in order to have a meaningful result.

I'd question their methodology as well.  If you're measuring system power consumption, then that includes other components, not just the video card.  Games tend to push a processor in addition to a video card, so you get some added power consumption there.  Using a Bloomfield/X58 setup that is horrible on power consumption will make that effect a lot worse than it needs to be.  Even if they somehow found an application that doesn't touch the processor at all, you're still adding in power supply inefficiencies.  That they don't account for that (which really isn't that hard to do) is rather shocking.  So they're probably seriously overestimating how much power the video cards use in their test.

And then you should consider that that's nowhere near a real max load.  Consider, for example, that reviews of the Radeon HD 6970 found that Metro 2033 was able to make PowerTune kick in now and then.  That's with a cap of 250 W, which means that Metro 2033 was able to pull 250 W from the video card alone.  The Guru 3D test finds power consumption of 209 W, and once you subtract off the reasons why they're overestimating things, they're probably seeing under 200 W of power consumption.  That's not a max load on the card.  It might be typical power consumption for a game where the video card is the bottleneck, but the T in TDP doesn't stand for "typical".

The reviewing site uses their own testing method, rather than benchmarks that the GPU manufacturers can game via drivers. The chart compares the cards tested by the site using their method. So, it's internally accurate, but may disagree with some of the values determined by other sites. :)

Originally posted by Hopscotch73

Quick question for Quizzical -

I'm in a similar boat to the OP (except my card is a 240), and I was already looking at the 560ti - came across a MSI N560GTX-Ti Twin Frozr II 2GD5/OC and was wondering if the 2GB vram makes it a better choice, it's only €20 more than the 1gb version, and while more is generally better in terms of VRam, I know it's not always the case, so I thought I'd ask the resident guru.

OP - sorry for slight derail!

What else do you have in your system?  If you don't have the case and power supply to handle it, sticking a much higher power video card in your system can fry things.

You don't really need more than 1 GB of video memory unless you're trying to use a resolution above 1920x1200.  If you want to play games at a resolution of 2560x1440 or larger, then you'd want a 2 GB card--but probably something faster than a GeForce GTX 560 Ti for that high of resolution.  As with system memory, video memory is largely a matter of, either you have enough or you don't.  If you have enough, then it doesn't matter if you have a little more than you need or ten times as much as you need.  The extra won't get used, anyway.

Cards that double the normal amount of video memory tend to be low volume parts, and that drives prices up.  It's very rare that they're a good value for the money, and I strongly suspect that you're saying, it's only €20 more than some other card that is also overpriced.  Where are you looking at buying a video card, anyway?

Originally posted by Quizzical

Maximum power consumption is tricky to test, because if AMD and Nvidia know that it's a power consumption test, they'll have their drivers clock the card back severely to look like they're lower power.  My view is that you need to give performance figures and power consumption from the same test in order to have a meaningful result.

I'd question their methodology as well.  If you're measuring system power consumption, then that includes other components, not just the video card.  Games tend to push a processor in addition to a video card, so you get some added power consumption there.  Using a Bloomfield/X58 setup that is horrible on power consumption will make that effect a lot worse than it needs to be.  Even if they somehow found an application that doesn't touch the processor at all, you're still adding in power supply inefficiencies.  That they don't account for that (which really isn't that hard to do) is rather shocking.  So they're probably seriously overestimating how much power the video cards use in their test.

And then you should consider that that's nowhere near a real max load.  Consider, for example, that reviews of the Radeon HD 6970 found that Metro 2033 was able to make PowerTune kick in now and then.  That's with a cap of 250 W, which means that Metro 2033 was able to pull 250 W from the video card alone.  The Guru 3D test finds power consumption of 209 W, and once you subtract off the reasons why they're overestimating things, they're probably seeing under 200 W of power consumption.  That's not a max load on the card.  It might be typical power consumption for a game where the video card is the bottleneck, but the T in TDP doesn't stand for "typical".

So?

There be a chart there, they all used the same method, at least if not 100% accurate shows a degree or trend of accuracy.

Nm, quizz is always correct, Quizz is always correct (I wonder if this will keep me from having having to respond to 3-5 paragraphs, i doubt it).

---

Originally posted by Jetrpg
So?
There be a chart there, they all used the same method, at least if not 100% accurate shows a degree or trend of accuracy.
Nm, quizz is always correct, Quizz is always correct (I wonder if this will keep me from having having to respond to 3-5 paragraphs, i doubt it).

---

Your better off looking at the TDP of the card - that's how much thermal power the card is rated to produce, and since there is a thing called the First Law of Thermodynamics, what comes out equal to what goes in. So that means the power drawn by the video card is at least proportional to the TDP rating of the waste heat produced by the card (minus the electricity that actually goes into producing work (which is amazingly little) and to run the cooling system, and conversion inefficiencies, etc. - which all together are a pretty small part of that number)

Now, I won't claim that TDP's are entirely accurate, nVidia in particular has been known to stretch the truth on these when it suits them - mainly because they have no way of actually capping a card to prevent it from going beyond it's rated TDP. And overclocks invalidate them, and you can get some really nice yield chips that will come in under TDP. But they should be close, and fairly representative, and make for a good comparison tool if nothing else.

But it's probably a better benchmark than someone sitting with a $20 Kill-o-Watt meter plugged into a computer trying to divine how much of that total load is due to the video card and how much is due to driver differences, or because there is a hard coded exception in the driver (*cough* Furmark) , or because one card can offload physics and etc...

I do agree - that one web site is internally consistent if it runs the same test, and just swaps out video cards. However, they can't possibly account for things like internal driver hardcodes - which are very common, and before PowerTune, was the only way that GPU manufacturers had to protect their cards against run-away power consumption. If you were to actually bench something like Furmark without the hard code, nVidia cards would run off the charts (possibly damaging themselves in the process), and AMD cards would just keep bumping their PowerTune settings. So while it's internally consistent, it can't account for things like this.

Ill take ACTUAL readings, scientific measurements (if not totally controlled), over manufacturer's Ratings anyday. Similarly, i can see the flop rate and its computing power of two videos cards , cpus, etc. And the highest isn't always the best preformer. Well By that rating it should be.

This isn't to say the rating is wrong. Only that i feel actual measurements seem more controlled/ accurate.

Tho i totally, believe the 570 has higher power consumption, just would have used sources to prove that , or tell them to google it some more.

Ps- edit- Power consuption is a joke, picking one or another gpu using that as a metric is silly.

Originally posted by Jetrpg

Ill take ACTUAL readings, scientific measurements (if not totally controlled), over manufacturer's Ratings anyday. Similarly, i can see the flop rate and its computing power of two videos cards , cpus, etc. And the highest isn't always the best preformer. Well By that rating it should be.

This isn't to say the rating is wrong. Only that i feel actual measurements seem more controlled/ accurate.

Tho i totally, believe the 570 has higher power consumption, just would have used sources to prove that , or tell them to google it some more.

Ps- edit- Power consuption is a joke, picking one or another gpu using that as a metric is silly.

Look at the chart in question.  It says "Calculated card TDP".  TDP doesn't mean "power used when playing a typical game".

TDP stands for Thermal Design Power.  It means, this is the most power that the card could plausibly use for thermally-significant lengths of time (1 minute is significant, but 1 ms is not).  TDP is a way for card manufacturers to tell people who build systems, this is the amount of power that you need to be able to deliver to the video card, and the amount of heat that you need to be able to dissipate from it.  If you can deliver this much power and dissipate this much heat (while keeping an ample supply of suitably cool air delivered the the card intake fan(s)), then we promise that the card will work safely in your system.

The amount of power that a card uses in an average game is irrelevant here.  If a given video card uses 100 W in game A, 150 W in game B, and 200 W in game C, then you might reasonably say that it uses 150 W in an average game.  But if the card is designed to be safe at 150 W and not at 200 W, then game C might well fry the card.  The TDP needs to be at least 200 W.

Which measurement is relevant depends on what you're trying to do.  If the question is what your electric bills are going to be, then TDP doesn't matter.  Average power consumption is what matters--and not just in games, either; idle power consumption actually plays a huge role here.  If you're trying to pick out a case and power supply, then average gaming power consumption doesn't matter.  If you want to avoid frying things, the TDP is what matters.

AMD cards with PowerTune have a simple way to give the TDP:  whatever the PowerTune cap is.  If a card has a PowerTune cap of 200 W, and some power virus shows up that would make the card pull 250 W, then PowerTune will throttle back the clock speeds so that the card still only uses 200 W.

Furthermore, the TDP is adjustable by the end-user, as you can change the PowerTune cap.  If you get a Radeon HD 7970 with its TDP of 250 W, and then set the PowerTune cap to 200 W, then the card now has a TDP of 200 W.  That won't reduce typical gaming power consumption by 50 W; in many games it won't have any effect at all, and even for games where it does, it might reduce typical power consumption by 5 or 10 W, not 50.

For cards that don't have anything analogous to PowerTune, the TDP is a number that the company has to pick.  For Fermi cards, the TDP that Nvidia listed was basically a lie, to try to cover up how bad the cards were on power consumption.  The generation before that, when Nvidia and AMD were competitive on performance per watt, Nvidia listed more honest TDPs that would be rather difficult to reach in real games.

"There be a chart there, they all used the same method, at least if not 100% accurate shows a degree or trend of accuracy."

That's why they play the games.  Well, some sites don't.  But that's why we look to Hard OCP for video card reviews.  If you can get higher performance with lower power consumption in real games (not just canned benchmark demos of real games), then that's not cheating benchmarks.  That's what video cards are supposed to try to do.

Nvidia 560 Gtx Ti either single or in SLI

---

Originally posted by Jetrpg
Ill take ACTUAL readings, scientific measurements (if not totally controlled), over manufacturer's Ratings anyday. Similarly, i can see the flop rate and its computing power of two videos cards , cpus, etc. And the highest isn't always the best preformer. Well By that rating it should be.
This isn't to say the rating is wrong. Only that i feel actual measurements seem more controlled/ accurate.
Tho i totally, believe the 570 has higher power consumption, just would have used sources to prove that , or tell them to google it some more.
Ps- edit- Power consuption is a joke, picking one or another gpu using that as a metric is silly.

---

My point is, these readings aren't even close to scientific.

At least with the TDP, if it's totally bogus, you can hold the company publishing that number responsible.

Would I pick a video card based on a power metric? All other things equal, or near equal, I absolutely would. Small form factor, or low noise, I absolutely would. Laptop - exclusively and over near all other factors. It's not just some meaningless statistic, it has real world ramifications, even in desktop computers. If the first wave of 480GTX's didn't prove that, the 590 GTX certainly did.

And if someone doesn't care to google (or research, as a scientific-minded person may put it) my statements to determine their validity - I don't care. You can believe me, or verify what I say, or just dismiss it. It doesn't affect the truth (or ignorance) of my statements.

I'll even, this time, include some links to "prove" my points:

http://hardocp.com/article/2010/09/30/my_quiet_galaxy_geforce_gtx_480_sli_build/

---

Running these 480 cards at full load for a period of 30 minutes has the ability to raise the temperature in my office a few degrees easily. I can certainly feel the 480 exhaust pouring out from under my desk while gaming. Actually I can even feel the heat with the box simply at a 2D desktop. Even though I have gotten the box a bit further from me, I am still thinking about building a little heat shield so it directs the heat behind my back rather than at my side.

---

http://www.youtube.com/watch?v=D-peJeY8zWA
(this one made me laugh out loud)

http://hardforum.com/showthread.php?t=1596174

http://www.guru3d.com/article/geforce-gtx-590-review/21

---

Wrapping up, you know I'm still a little flabbergasted as to why NVIDIA clocked the card as low as they did. Likely they are ying-yang with the fact that this card performance roughly equal slash close to R6990 performance and simply opted for low noise and a fashionable power consumption.

---