Is the 7950 or 670 truly faster

[The Alias]

I think you are being a bit generous to the 7950 there.

no not really 7950s at that speed are blazing fast and most 670s boost around that at stock anyway sooo yeah .. no

 

[Zanovar]

no not really 7950s at that speed are blazing fast and most 670s boost around that at stock anyway sooo yeah .. no

its close.Im curios what do you think most 670s boost to anyways?

 

[ICDP]

My GTX 670 boosts to 1240mhz at stock, does anyone know how that compares to a 7950?

My tests of HD 7970, HD 7950 and GTX 680 showed the following trends, taking the HD 7950 as a baseline. All results were taken at 2560x1600 resolution.

A GTX680 needs to be ~75MHz faster than a HD 7950 to match it for performance.

HD 7970 needs to be ~75MHz slower than a HD 7950 for the 7950 to match it.

For reference, a GTX680 needs to be ~150MHz faster than a HD 7970 to match it for performance.

GTX680 benefits far more from a VRAM overcock than a core overclock. The reverse is true for a HD 7970 and to a slightly lesser extent the 7950.

Of course is assuming VRAM is overclocked also. The GTX 680 needs a hefty VRAM boost to keep up at this resolution when all 3 are overclocked.

 

[Zanovar]

My tests of HD 7970, HD 7950 and GTX 680 showed the following trends, taking the HD 7950 as a baseline. All results were taken at 2560x1600 resolution.

A GTX680 needs to be ~75MHz faster than a HD 7950 to match it for performance.

HD 7970 needs to be ~75MHz slower than a HD 7950 for the 7950 to match it.

For reference, a GTX680 needs to be ~150MHz faster than a HD 7970 to match it for performance.

GTX680 benefits far more from a VRAM overcock than a core overclock. The reverse is true for a HD 7970 and to a slightly lesser extent the 7950.

Of course is assuming VRAM is overclocked also. The GTX 680 needs a hefty VRAM boost to keep up at this resolution when all 3 are overclocked.

Good stuff,wheres me graphs ,only messing.

 

[ICDP]

I think some benchmark results would actually help, it's better than some words on a forum

All taken using Heaven 3.0.

HD 7950 1230/1700

GTX 680 1228 core and +554 VRAM

HD 7970 1200/1550

Honestly, the difference in speed between a GTX 670, GTX680, HD 7970 and HD 7950 depends on the overclock and the game you are playing. The only reason I recommend AMD over Nvidia at this point in time is price. Similar speed, lower price, better games bundle.

 

[notty22]

Sorry one synthetic does not represent a cross-section of games. What is the reason that since you have been posting every discussion is brought back to you having the hardware, and your eyes/results.
Do you still have a 680/ 7970 and 7950 ?

Don't forget the very large price premium you pay for the slower Nvidia products. Things really have turned on their head since the GTX 680 release

 

[amenx]

My tests of HD 7970, HD 7950 and GTX 680 showed the following trends, taking the HD 7950 as a baseline. All results were taken at 2560x1600 resolution.

A GTX680 needs to be ~75MHz faster than a HD 7950 to match it for performance.

A 680 at stock rapes a 7950 at stock, even at 2560x1600 (overall). I am sure one can cherry pick benches that show otherwise, but overall, the 680 should be clearly ahead of a 7950.

I also see in the Heaven benches you have the 7950 at 1230/1700. Quite a hefty OC for a 7950. Looks like you won the silicone lottery. How representative is that of typical 7950s? I hear the norm is around 1100.

 

[OILFIELDTRASH]

A 680 at stock rapes a 7950 at stock, even at 2560x1600 (overall). I am sure one can cherry pick benches that show otherwise, but overall, the 680 should be clearly ahead of a 7950.

I also see in the Heaven benches you have the 7950 at 1230/1700. Quite a hefty OC for a 7950. Looks like you won the silicone lottery. How representative is that of typical 7950s? I hear the norm is around 1100.

Comparing stock 680 to stock 7950 really isn't fair. Nvidia nearly squeezed every drop of performance from the 680 with their version on boost clocks. 7950 is sandbagging majorly. I'd be shocked if the norm for the 7950 wasn't closer to 1200mhz. Mine is running low volts and is still cruising along at 1122 without even trying. I wouldn't be surprised to see 1250mhz with more volts.
The only rape going on is Nvidia pricing and fleecing of it's customer.

 

[aaksheytalwar]

1200mhz is far from the norm for a 7950 under realistic conditions. 1050-1150 is more like it with few going beyond 1150 at reasonable long term temps etc.

 

[amenx]

Comparing stock 680 to stock 7950 really isn't fair.

I was not the one making that point, the poster I quoted was and I was simply taking him up on it:

"A GTX680 needs to be ~75MHz faster than a HD 7950 to match it for performance."

Do you agree with that?

 

[Eureka]

Comparing stock 680 to stock 7950 really isn't fair. Nvidia nearly squeezed every drop of performance from the 680 with their version on boost clocks. 7950 is sandbagging majorly. I'd be shocked if the norm for the 7950 wasn't closer to 1200mhz. Mine is running low volts and is still cruising along at 1122 without even trying. I wouldn't be surprised to see 1250mhz with more volts.
The only rape going on is Nvidia pricing and fleecing of it's customer.

FYI, it isn't. The norm for a 7950 is closer to 1050-1100. There's a silicon limit that starts around 1150.

Even though I like my 7950, stop spreading this overclock BS. Not every chip is capable of hitting those high numbers.

 

[Termie]

ICDP - thanks, this is really useful. Adding a GTX670 to the mix, clock-for-clock with your 680.

GTX670 1228 core, 7114 VRAM

Of note - PCIe 3.0 on my rig adds about 3% to the score, so the 680 on your system would be more like 5-6% faster clock-for-clock.

I think some benchmark results would actually help, it's better than some words on a forum

All taken using Heaven 3.0.

HD 7950 1230/1700

GTX 680 1228 core and +554 VRAM

HD 7970 1200/1550

Honestly, the difference in speed between a GTX 670, GTX680, HD 7970 and HD 7950 depends on the overclock and the game you are playing. The only reason I recommend AMD over Nvidia at this point in time is price. Similar speed, lower price, better games bundle.

 

[ICDP]

Sorry one synthetic does not represent a cross-section of games. What is the reason that since you have been posting every discussion is brought back to you having the hardware, and your eyes/results.
Do you still have a 680/ 7970 and 7950 ?

You just won't accept that the HD 7950 is a great card. No matter how much evidence is shown you still insist the HD 7950 is a stuttery mess and that a GTX 660Ti is its equal. Do you seriously think I went out of my way to purchase a GTX680, just so I could lie and claim a HD 7950 is its equal when overclocked.

I build PCs for a living and as a result I have owned or tested 4x GTX 680s, 6x HD 7970s and 2x HD 7950s in my system over the course of this year. I sold my HD 7970 and GTX 680 two weeks ago to cut my losses on both. Considering they are roughly matched at 2560x1600 by my HD 7950 which cost me considerably less I felt I may as well sell both before Titan was launched.

Here are a couple of game benchmarks.

Hitman Absolution: 7950 at 1175/1550

Hitman Absolution: GTX 680 1202 core and +554VRAM

Far Cry 3
2560x1600
All settings at max/ultra in game, 2x MSAA. The 7950 is slightly ahead on average FPS but again the GTX 680 gave better minimums, though in gameplay these minimums are not noticeable

Far Cry 3: 7950 at 1175/1550

Far Cry 3: GTX 680 1202 core and +554VRAM

 

[notty22]

Who said it wasn't a good card? You do this for a living hey?
A 680 o/c's also.

http://www.techpowerup.com/reviews/MSI/HD_7950_Twin_Frozr_III/29.html
7950_Twin_Frozr_III
Overclocking

The overclocks listed in this section were achieved with the default fan and voltage settings as defined in the VGA BIOS. Please note that every single sample overclocks differently, that's why our results here can only serve as a guideline for what you can expect from your card.

Maximum stable clocks of our card are 1020 MHz core (16% overclock) and 1650 MHz Memory (32% overclock).

GPU overclocking works well, but the final overclock seems a bit below what we have seen on other HD 7950 cards (AMD ref: 1085 MHz, PowerColor PCS+: 1145 MHz).

Memory overclocking works great, and is exactly in the range of typical overclocks for this memory.

http://www.techpowerup.com/reviews/MSI/GTX_680_Lightning/31.html
MSI/GTX_680_Lightning

Maximum stable clocks of our card are 1225 MHz core (10% overclock) and 1780 MHz Memory (19% overclock).
Using these clock frequencies we ran a quick test of Battlefield 3 to evaluate the gains from overclocking.

Actual 3D performance gained from overclocking is 13.4%.

54 VS 86 ? This is atypical of what's posted lately O/C vs stock, example.

 

[wand3r3r]

You just won't accept that the HD 7950 is a great card. No matter how much evidence is shown you still insist the HD 7950 is a stuttery mess and that a GTX 660Ti is its equal. Do you seriously think I went out of my way to purchase a GTX680, just so I could lie and claim a HD 7950 is its equal when overclocked.

I build PCs for a living and as a result I have owned or tested 4x GTX 680s, 6x HD 7970s and 2x HD 7950s in my system over the course of this year. I sold my HD 7970 and GTX 680 two weeks ago to cut my losses on both. Considering they are roughly matched at 2560x1600 by my HD 7950 which cost me considerably less I felt I may as well sell both before Titan was launched.

Here are a couple of game benchmarks.

Evidence is meaningless to team green. The only valid evidence is when it's leaning in their favor.

^see the post above if it doesn't get edited.

 

[notty22]

Evidence is meaningless to team green. The only valid evidence is when it's leaning in their favor.

^see the post above if it doesn't get edited.

Typical of a hypocrite, you just edited your post. I'll call you the commentator. Who only whines about Teams.

 

[ICDP]

A 680 at stock rapes a 7950 at stock, even at 2560x1600 (overall). I am sure one can cherry pick benches that show otherwise, but overall, the 680 should be clearly ahead of a 7950.

Absolutely true, the GTX 680 does rape a 7950 when both are at stock. Though because the HD 7950 has a much higher headroom for overclocking they come very close when both are at max overclock. That is the beauty of the HD 7950, top-end performance for mid range prices.

I also see in the Heaven benches you have the 7950 at 1230/1700. Quite a hefty OC for a 7950. Looks like you won the silicone lottery. How representative is that of typical 7950s? I hear the norm is around 1100.

I have owned or tested 2x HD 7950s in the past year.

1x MSI reference. Max stable OC was 1170/1600

1x MSI TF3. Max stable OC 1230/1700 (the one I tested with for this thread. From what I can tell the norm is ~1150 and at that speed it is faster than a GTX 680 at stock.

For comparison I have owned or tested 4x GTX 680s in the past year.

1x EVGA reference 680 2GB. Max stable OC was 1310 and + 470 VRAM. An excellent OC but it did run hot at these settings.

1x Gainward reference 680 2GB. Max stable OC was 1180 and + 180 VRAM
An absolute stinker of a card as far as overclocking. Thankfully the person who I was buying it for had no interest in overclocking.

1x KFA2 reference GTX 680 2GB. Max stable OC of 1238 and +350 VRAM.
A slightly above average OC, nothing special all things considered.

1x MSI Lightning GTX 680 2GB. Max stable OC of 1356 and +554 VRAM.
Not a great overclock for such an expensive card. It still overclocked beyond the average GTX 680 though.

So HD 7950 will get ~1150 with a voltage tweak.
GTX 680 will get ~1270.

 

[wand3r3r]

Typical of a hypocrite, you just edited your post. I'll call you the commentator. Who only whines about Teams.

Yeah I added another example (when I saw the argument going even further in left field), I didn't need to it was already evident in the thread. How hypocritical.

A GTX680 needs to be ~75MHz faster than a HD 7950 to match it for performance.

HD 7970 needs to be ~75MHz slower than a HD 7950 for the 7950 to match it.

For reference, a GTX680 needs to be ~150MHz faster than a HD 7970 to match it for performance.

Your response to that is what? First you whine about a single synthetic benchmark, then you bring in the 680 lightning vs. a weak 7950. Talk about cherry picking and going off the point, or you haven't made any clear argument in any case.

 

[ICDP]

Who said it wasn't a good card? You do this for a living hey?
A 680 o/c's also.
http://www.techpowerup.com/reviews/MSI/GTX_680_Lightning/31.html
MSI/GTX_680_Lightning

54 VS 86 ? This is atypical of what's posted lately O/C vs stock, example.

So you linked to a review of an MSI GTX 680 Lightning and an MSI TF3 7950. That is one hell of a coincidence because the results of my tests were taken from using the exact same cards.

Some points worth noting.

1. Neither review touched voltage in either card.
2. Even the reviewer notes that the 7950 overlcock achieved is quite low compared to other 7950s he tested.
3. An overclock of 1020 on a HD 7950 is actually a decent result without touching voltage.
4. A GTX 680 Lightning is around £200 more expensive that a TF3 7950.

 

[ICDP]

ICDP - thanks, this is really useful. Adding a GTX670 to the mix, clock-for-clock with your 680.

GTX670 1228 core, 7114 VRAM

Of note - PCIe 3.0 on my rig adds about 3% to the score, so the 680 on your system would be more like 5-6% faster clock-for-clock.

This just goes to show how much of a bargain the GTX 670 is compared to a GTX 680. When the GTX 670 was released it was by far the best bargain you could get in a price/perf comparison. Then again I suspect you knew this already.

There really is very little to choose between the HD 7950, 7970, GTX670 and GTX 680 when all are overclocked. This is why I cut my losses by keeping my HD 7950 and selling my GTX 680 Lightning and HD 7970.

 

[wand3r3r]

So you linked to a review of an MSI GTX 680 Lightning and an MSI TF3 7950. That is one hell of a coincidence because the results of my tests were taken from using the exact same cards.

Some points worth noting.

1. Neither review touched voltage in either card.
2. Even the reviewer notes that the 7950 overlcock achieved is quite low compared to other 7950s he tested.
3. An overclock of 1020 on a HD 7950 is actually a decent result without touching voltage.
4. A GTX 680 Lightning is around £200 more expensive that a TF3 7950.

In the lightning review they also state this.
[*] The maximum clock of 1225 MHz is higher than on any other GTX 680 we reviewed so far.

Lol, these aren't helping much.

 

[amenx]

Evidence is meaningless to team green. The only valid evidence is when it's leaning in their favor.

Team green? Ah... I see where you're coming from. Doesnt help you're argument other than to boldly state "team red" rules and one persons tests beats multiple review sites tests. Never mind the BF3 example, which sure, later driver optimizations have improved AMD cards considerably with. But Hitman is not representative of most games, its one of the rare games that a 7950 does indeed beat a 680, and in which by the same token a few games can be found where a 660ti can beat a 7950. I could basically make the same argument ICDP is making with a few selected games that the 660ti is faster than a 7950.

 

[ICDP]

Team green? Ah... I see where you're coming from. Doesnt help you're argument other than to boldly state "team red" rules and one persons tests beats multiple review sites tests. Never mind the BF3 example, which sure, later driver optimizations have improved AMD cards considerably with. But Hitman is not representative of most games, its one of the rare games that a 7950 does indeed beat a 680, and in which by the same token a few games can be found where a 660ti can beat a 7950. I could basically make the same argument ICDP is making with a few selected games that the 660ti is faster than a 7950.

Nice move, attempt to shoot down my tests by claiming they are not representative due to one game. How about the Heaven 3.0 benches, or the Far Cry 3 benchmarks, are they worthless also?

With both cards at stock speeds a GTX 680 is far faster than a HD 7950 boost edition for Hitman Absolution.
http://www.hardocp.com/article/2012/12/20/hitman_absolution_performance_iq_review/4#.UTOiNldi5I0

My charts show how they perform when overclocked to similar clocks. Far Cry3 on the other hand performs better on Nvidia hardware despite being a Gaming Evolved game.

Here we see a GTX 680 beat a HD 7970 GE in Far Cry 3 at 2560x1440. You can rest assured that the 7950 at stock speeds is nowhere close. So despite your bold claims, the tests are indeed representative.
http://www.anandtech.com/show/6774/nvidias-geforce-gtx-titan-part-2-titans-performance-unveiled/11

 

[notty22]

The OP asked about 7950 or gtx 670.

Here is a recent review, declaring the 660ti a better option than the 7950. But what do they know.

AMD Radeon HD 7950 Boost vs. Nvidia GeForce GTX 660 Ti: frametimes review

They cost about the same, but which one is the fastest?

 

[-Slacker-]

Stumbled across this review, reading Titan reviews the other day.
AMD Radeon HD 7950 Boost vs. Nvidia GeForce GTX 660 Ti: frametimes

These games were tested.

Assassin's Creeed III

Max Payne 3

Battlefield 3
Far Cry 3
Hitman: Absolution

Ah yes, who could forget uk.hardware.info - that super popular benchmarking website that is totally not obscure and everyone has heard of. I can imagine that you first meant to post results form lesser known sites like anandtech, hardocp, techpowerup etc but then figured that would be redundant as the fabled uk.hardware.info would cover all our curiosities on all things tech.

Just our luck, as I see that the review you submitted dispels many misconceptions that the other sham sites would have us believe, like the oveclocked hd 7950 not being slower than the stock gtx 660ti. Isn't coincidence grand?