GTX760 Review Thread

Page 4 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Feb 19, 2009
10,457
10
76
Bitmining is more stressful on radeons than gaming, I can attest that it lowered my stable gaming clocks by around 50mhz, as in, it would crash randomly after a few hours, or after a day etc.

Also note, 1.3ghz is near the peak for Kepler, outside of bios mods and extra voltage. 1.26ghz ingame boost is the average so far across reviews. 7950 seems to peak out at 1.25ghz, and the average across many forum threads and hwbot is at 1.15ghz. 7970 has a 50mhz extra headroom. Surprisingly, both vastly different architectures (Kepler vs GCN), yet their peak clocks tend to fall within a similar range.
 

24601

Golden Member
Jun 10, 2007
1,683
39
86
Bitmining is more stressful on radeons than gaming, I can attest that it lowered my stable gaming clocks by around 50mhz, as in, it would crash randomly after a few hours, or after a day etc.

Also note, 1.3ghz is near the peak for Kepler, outside of bios mods and extra voltage. 1.26ghz ingame boost is the average so far across reviews. 7950 seems to peak out at 1.25ghz, and the average across many forum threads and hwbot is at 1.15ghz. 7970 has a 50mhz extra headroom. Surprisingly, both vastly different architectures (Kepler vs GCN), yet their peak clocks tend to fall within a similar range.

I don't count bios mod as cheating in overclocking.

As long as it's stable I'd count it as valid, and basically saying people who don't bios mod 670/680/760/770/780/Titan aren't really enthusiasts, since there are pretty huge >>stable<< gains that are left on the table that you can pretty much automatically get.

My big problem for 7950 and 7970 overclocking is that the shipping thermal solutions are just barely adequate.

I want a Titan cooler for my Sapphire 7970 and 7950 so that my crossfire rig doesn't stupidly overheat instantly.
 
Last edited:

Elfear

Diamond Member
May 30, 2004
7,115
690
126
People submit at maximum possible boot clocks, not maximum possible stable clocks.

Example from my Sapphire 7970 in my sig.

My maximum possible boot clocks are 1300 core 1900 ram.

My maximum possible stable clocks at manufacturer spec (as in 0% artifacts/crashes anything weird in every single stress test I can do) is 1100 core 1700 ram, 1050 core 1800 ram.

These are very important and often glossed over differences.

That's why I don't brag about a 5ghz processor since I've only done full validation for the 4.6ghz in my sig.


You also forget that people don't advertise bad overclocks on an overclock forum.
That's just culturally discouraged.

I didn't buy my hardware to get arbitrary high scores on benchmarks with my whole screen artifacting / unstable.

Right, but the point I'm trying to make is that the chance of getting 1100 on a 7950 is not the same as getting 1300-1400 on a 760 which you stated. If you want to say that the submissions for HWBot are suicide clocks than you also have to say the same thing about the 760s. If you think a 7950@1100 is a golden sample than you must believe a 760@1210 is a golden sample as well.

I'm not sure why you think 1100 is improbable on a 7950. Go look at the 7950 overclocking thread. Plenty of people running 1200+ for gaming. I know it's not the same thing but my reference XFX 7970 is game stable at 1225Mhz using the stock cooler (50% fan speed). The 7000 series cards have proven to be great overclockers.
 

24601

Golden Member
Jun 10, 2007
1,683
39
86
Right, but the point I'm trying to make is that the chance of getting 1100 on a 7950 is not the same as getting 1300-1400 on a 760 which you stated. If you want to say that the submissions for HWBot are suicide clocks than you also have to say the same thing about the 760s. If you think a 7950@1100 is a golden sample than you must believe a 760@1210 is a golden sample as well.

I'm not sure why you think 1100 is improbable on a 7950. Go look at the 7950 overclocking thread. Plenty of people running 1200+ for gaming. I know it's not the same thing but my reference XFX 7970 is game stable at 1225Mhz using the stock cooler (50% fan speed). The 7000 series cards have proven to be great overclockers.

I'm saying the people doing proper Nvidia Geforce 6xx and 7xx overclocking (Bios Modding) are the definite minority, heavily skewing the results downwards. If you look at the 770 thread, you'll see that 1400 boost clocks are basically standard. EVGA also said in the 760 thread that most 760s they tested get 1300 boost clocks.

The reason for the huge disparity between that and the stable 7970 and 7950 clocks is that Tahiti has an extremely weak ROP/IMC implementation, which requires about .1-.2 extra voltage for vcore to keep them stable under maximum load.

What this means is you can either get high core clocks OR high memory clocks (if you have well binned GDDR5)

On top of that, many 7950s have low binned GDDR5 to start with, making that choice for the end user.

On top of that, many 7950s have extremely low ASIC quality, mostly 40-50%, with some models going 60-70% maximum.

Only select 7950 models have both high binned GDDR5 as well as high binned ASICs, as well as good voltage regulation, as well as good cooling solution.

Sadly the MSI TwinFrozr III >>can<< have high ASIC quality (about 50% of them do), and have ok voltage regulation, however have poor VRM cooling, and barely adequate general cooling performance (The TwinFrozr III is really a 230w cooler).

The only card I have seen so far in 7950 and 7970 that have all their ducks lined up is the HIS IceQ X^2 7950 and 7970.
The problem is that the 7950 model of that is extremely overpriced and the 7970 model of that is not price competitive with other models either.

And I make sure to specify manufacturer level of stress testing, as in every single thing you could possibly do to stress the card, no holds barred.

An unstable cards is infinitely less useful than a slightly slower, 100% stable card in actual use.

People glossing over these very important facts are painting an extremely skewed picture of the situation.

In comparison, the Windforce 3x REV2.0 760 has superior cooling to every single 7950 and 7970 currently purchasable for comparable price (and arguably superior cooling to every single one of them using conventional air cooling). This allows the bottleneck of clocks to be shifted back to the artificial TDP limit as well as the artificial Voltage limit, both easily to circumvent with simple Bios Mod tools.

Here's an anecdote from myself.

I can do 1100 core 1700 gddr5 at 1.050 set voltage on my Sapphire Dual-X 7970 and be stable in every non ROP intensive task. However, once ROPs are stressed to the max, the card requires 1.156 set voltage.

Another anecdote.

I can do 1125 core 1700 gddr5 at 1.188 set voltage on my Sapphire Dual-X 7950 with moderate ROP load, however, I need to set voltage at 1.256v for maximum stressed ROPs. However, this voltage produces far too high temperatures at maximum GPU, ROP, and VRM stress, making 1125 core 1700 gddr5 not truly stable.

This is why I have been harping on AMD getting something rivaling the Titan cooling solution and mirroring Nvidia's Greenlight program for cooling solutions.

If I had Titan coolers on my Sapphire 7970 and 7950 I would be able to clock far higher 100% stable at 100% GPU/ROP/VRM load as temperatures in Crossfire configuration would not skyrocket as high as it does because of the fact that the stock blower configuration of 7970 and 7950 is atrociously bad as well as the fact that to get good cooling you are required to go open air cooling, which is terrible if you have 3+ cards since no matter what you do, you will only have 1 slot at most between each card.

This also has resulted in the fact that I literally cannot do quadfire with my 4 cards, since each card (except for the bottom one) would overheat at any reasonable load even at manufacturer default settings.

Note this is with possibly the best possible case cooling solution on the market currently for air cooling.
I have every single fan slot in my Antec 1100 case occupied with Antec Tri-cool fans at high setting (except for the 2 recirculating locations in the interior) in the configuration that tested as lowest temperatures for GPUs (Took me a while).
 
Last edited:

Elfear

Diamond Member
May 30, 2004
7,115
690
126
I'm saying the people doing proper Nvidia Geforce 6xx and 7xx overclocking (Bios Modding) are the definite minority, heavily skewing the results downwards. If you look at the 770 thread, you'll see that 1400 boost clocks are basically standard. EVGA also said in the 760 thread that most 760s they tested get 1300 boost clocks.

Your evidence is anecdotal compared to the 114 submissions on HWBot for the 770 which put the average OC at 1244/1807. HWBot is made up of enthusiasts who will take their cards to the max.



If I had Titan coolers on my Sapphire 7970 and 7950 I would be able to clock far higher 100% stable at 100% GPU/ROP/VRM load as temperatures in Crossfire configuration would not skyrocket as high as it does because of the fact that the stock blower configuration of 7970 and 7950 is atrociously bad as well as the fact that to get good cooling you are required to go open air cooling, which is terrible if you have 3+ cards since no matter what you do, you will only have 1 slot at most between each card.

This also has resulted in the fact that I literally cannot do quadfire with my 4 cards, since each card (except for the bottom one) would overheat at any reasonable load even at manufacturer default settings.

Note this is with possibly the best possible case cooling solution on the market currently for air cooling.
I have every single fan slot in my Antec 1100 case occupied with Antec Tri-cool fans at high setting (except for the 2 recirculating locations in the interior) in the configuration that tested as lowest temperatures for GPUs (Took me a while).
I'm not arguing that your experience with your Tahiti cards isn't what it is but many have had good overclocking luck with less expensive cards than the Twinfrozer or Windforce models. The fourth card in my quadfire setup is a Dual-X Sapphire 7970. While mining away at 1050Mhz temps never go above 70C (fan@60% is impressively quiet) and it is a couple mm from the side of my case.

I agree the reference cooler is very loud when you start to crank up the fan speed for overclocking. Headphones or an aftermarket cooler are a must for overclockers.
 

Face2Face

Diamond Member
Jun 6, 2001
4,100
215
106
On top of that, many 7950s have low binned GDDR5 to start with, making that choice for the end user.

On top of that, many 7950s have extremely low ASIC quality, mostly 40-50%, with some models going 60-70% maximum.

Only select 7950 models have both high binned GDDR5 as well as high binned ASICs, as well as good voltage regulation, as well as good cooling solution.

Sadly the MSI TwinFrozr III >>can<< have high ASIC quality (about 50% of them do), and have ok voltage regulation, however have poor VRM cooling, and barely adequate general cooling performance (The TwinFrozr III is really a 230w cooler).

Very true when It comes to newer 7950's. However there is one thing I would like to point out on the MSI TF3's. While the VRM cooling on all TF3's is pretty horrible, not all cards have many issues. From talking with other people with a cards like mine, A TF3 V2 (6+8pin) 89.1% ASIC, they are all able to achieve high clocks (1200Mhz +) at lower voltage (1.150-1.175v) but at the cost of high core temps and high vrms temps. Newer lower ASIC quality TF3's with either 6+6 or 6+8 run much cooler and same clocks with 1.25v. Maybe Balla can chime in and show us his vrm temps @ 1200Mhz core playing Crysis 3? But from what I have seen across the web it's not an issue for every 7950 TF3 owner.
 
Last edited:

SilverBack

Golden Member
Oct 10, 1999
1,622
0
0
I just purchased an EVGA 760 4 GB. for $279
I multibox 5 characters in WoW , Rift and 4 in SWTOR.
This is absolutely the best bang for the buck on a multibox set up.
 

Final8ty

Golden Member
Jun 13, 2007
1,172
13
81
Originally Posted by Communism View Post


On top of that, many 7950s have low binned GDDR5 to start with, making that choice for the end user.

On top of that, many 7950s have extremely low ASIC quality, mostly 40-50%, with some models going 60-70% maximum.

Only select 7950 models have both high binned GDDR5 as well as high binned ASICs, as well as good voltage regulation, as well as good cooling solution.

Sadly the MSI TwinFrozr III >>can<< have high ASIC quality (about 50% of them do), and have ok voltage regulation, however have poor VRM cooling, and barely adequate general cooling performance (The TwinFrozr III is really a 230w cooler).

ASIC 56,60,73.

MSI HD 7950 OC BE
The cheapest 7950's that i could get, £207
http://www.overclockers.co.uk/showproduct.php?prodid=GX-203-MS&groupid=701&catid=56&subcat=1673 price has gone up.

- Core Clock: 830MHz
- Boost Clock: 840MHz
- Memory: 3072MB GDDR5
- Memory Clock: 5000MHz (Effective)

ALL do 1065/1465 on stock volts 1.170 and currently running at 1100/1465 on 1.2volts 60% fan and temps between 60-70c in tritfire with no spacing, had to put a piece of card between them.
 
Last edited:

BallaTheFeared

Diamond Member
Nov 15, 2010
8,115
0
71
Very true when It comes to newer 7950's. However there is one thing I would like to point out on the MSI TF3's. While the VRM cooling on all TF3's is pretty horrible, not all cards have many issues. From talking with other people with a cards like mine, A TF3 V2 (6+8pin) 89.1% ASIC, they are all able to achieve high clocks (1200Mhz +) at lower voltage (1.150-1.175v) but at the cost of high core temps and high vrms temps. Newer lower ASIC quality TF3's with either 6+6 or 6+8 run much cooler and same clocks with 1.25v. Maybe Balla can chime in and show us his vrm temps @ 1200Mhz core playing Crysis 3? But from what I have seen across the web it's not an issue for every 7950 TF3 owner.

I have a 8+6 which runs hotter than my 6+6, both are TF3 cards though.

In CF open bench with a fan on them to help airflow (not ideal).

Actually been playing Metro Last Light, so that is what you get, and not at 1200 >.<

1150 core 1250mv, 1600MHz memory 1681mv, vysnc on, but it doesn't really matter - it's on "Ultra" with 2xSSAA and 59 FoV. Basically I need this, lol... I have drops into the mid 40's during actual gameplay! 20nm can't come soon enough, need more powa!

Fan set to auto, was feeling lazy reinstalled my OS a few times since I got my SSD back... Anyways!



VRMs peak 73C/72C for the bottom card and 73C/65C for the top card (has cpu fan blowing on it).

i5 is of course at 4.6GHz 1.155v

This was over an hour of playing, and I should note outside of replacing the stock TIM on the gpu core itself I haven't fiddled with the memory or vrm cooling or it's pads.

Power was in flux around 470w at the wall I would say "on average" though tbh I never looked all that much

Rawr.
 
Last edited:

f1sherman

Platinum Member
Apr 5, 2011
2,243
1
0
I'm saying the people doing proper Nvidia Geforce 6xx and 7xx overclocking (Bios Modding) are the definite minority, heavily skewing the results downwards. If you look at the 770 thread, you'll see that 1400 boost clocks are basically standard. EVGA also said in the 760 thread that most 760s they tested get 1300 boost clocks.

The reason for the huge disparity between that and the stable 7970 and 7950 clocks is that Tahiti has an extremely weak ROP/IMC implementation, which requires about .1-.2 extra voltage for vcore to keep them stable under maximum load.

What this means is you can either get high core clocks OR high memory clocks (if you have well binned GDDR5)

On top of that, many 7950s have low binned GDDR5 to start with, making that choice for the end user.

On top of that, many 7950s have extremely low ASIC quality, mostly 40-50%, with some models going 60-70% maximum.

Only select 7950 models have both high binned GDDR5 as well as high binned ASICs, as well as good voltage regulation, as well as good cooling solution.

Sadly the MSI TwinFrozr III >>can<< have high ASIC quality (about 50% of them do), and have ok voltage regulation, however have poor VRM cooling, and barely adequate general cooling performance (The TwinFrozr III is really a 230w cooler).

The only card I have seen so far in 7950 and 7970 that have all their ducks lined up is the HIS IceQ X^2 7950 and 7970.
The problem is that the 7950 model of that is extremely overpriced and the 7970 model of that is not price competitive with other models either.

And I make sure to specify manufacturer level of stress testing, as in every single thing you could possibly do to stress the card, no holds barred.

An unstable cards is infinitely less useful than a slightly slower, 100% stable card in actual use.

People glossing over these very important facts are painting an extremely skewed picture of the situation.

In comparison, the Windforce 3x REV2.0 760 has superior cooling to every single 7950 and 7970 currently purchasable for comparable price (and arguably superior cooling to every single one of them using conventional air cooling). This allows the bottleneck of clocks to be shifted back to the artificial TDP limit as well as the artificial Voltage limit, both easily to circumvent with simple Bios Mod tools.

Here's an anecdote from myself.

I can do 1100 core 1700 gddr5 at 1.050 set voltage on my Sapphire Dual-X 7970 and be stable in every non ROP intensive task. However, once ROPs are stressed to the max, the card requires 1.156 set voltage.

Another anecdote.

I can do 1125 core 1700 gddr5 at 1.188 set voltage on my Sapphire Dual-X 7950 with moderate ROP load, however, I need to set voltage at 1.256v for maximum stressed ROPs. However, this voltage produces far too high temperatures at maximum GPU, ROP, and VRM stress, making 1125 core 1700 gddr5 not truly stable.

This is why I have been harping on AMD getting something rivaling the Titan cooling solution and mirroring Nvidia's Greenlight program for cooling solutions.

If I had Titan coolers on my Sapphire 7970 and 7950 I would be able to clock far higher 100% stable at 100% GPU/ROP/VRM load as temperatures in Crossfire configuration would not skyrocket as high as it does because of the fact that the stock blower configuration of 7970 and 7950 is atrociously bad as well as the fact that to get good cooling you are required to go open air cooling, which is terrible if you have 3+ cards since no matter what you do, you will only have 1 slot at most between each card.

This also has resulted in the fact that I literally cannot do quadfire with my 4 cards, since each card (except for the bottom one) would overheat at any reasonable load even at manufacturer default settings.

Note this is with possibly the best possible case cooling solution on the market currently for air cooling.
I have every single fan slot in my Antec 1100 case occupied with Antec Tri-cool fans at high setting (except for the 2 recirculating locations in the interior) in the configuration that tested as lowest temperatures for GPUs (Took me a while).

Great post!

But... After doing some research, I don't think Windforce rev2.0 heatsink on 760 is in the same class like Windforce GTX 780.
I don't think that particular 760 revision has been reviewed yet.


This is the cooler I want:

http://www.gigabyte.com/products/product-page.aspx?pid=4621#ov


but they are shipping this one with 760 rev 2.0

http://www.gigabyte.com/products/product-page.aspx?pid=4663#ov



Giga Windforce vs EVGA ACX vs Asus DC2 vs MSI TF4 vs Inno3D iChill... hmm.
 
Last edited:

24601

Golden Member
Jun 10, 2007
1,683
39
86
Great post!

But... After doing some research, I don't think Windforce rev2.0 heatsink on 760 is in the same class like Windforce GTX 780.
I don't think that particular 760 revision has been reviewed yet.


This is the cooler I want:

http://www.gigabyte.com/products/product-page.aspx?pid=4621#ov


but they are shipping this one with 760 rev 2.0

http://www.gigabyte.com/products/product-page.aspx?pid=4663#ov



Giga Windforce vs EVGA ACX vs Asus DC2 vs MSI TF4 vs Inno3D iChill... hmm.

Yes, the one on the 760 is not the 450w specification one.

It looks like exactly the same fin pitch and layout and identical fans and almost identical frame. The only difference I see is the probably thinner diameter heat-pipes and the heatsink array not extending as far towards the long side of the card, likely due to the PCB itself being far shorter than the 770/780 PCB.

However, it is still the best you can get for not much added to the cost of the card unless you are getting a 770 or 780 (The ones with the 450w specification Windforce 3x).

Reviewers have tested and the Windforce 3x Rev2.0 on the 760 is superior to the EVGA ACX, Asus DC2, and MSI TF4 for certain.

You'll have to find your own comparison review for the Inno3D iChill.
 
Last edited:

f1sherman

Platinum Member
Apr 5, 2011
2,243
1
0
Last edited:

24601

Golden Member
Jun 10, 2007
1,683
39
86
Well it has been double confirmed

Inno3D iChill cooler is just crazy:

http://uk.hardware.info/reviews/452...egment-cooling-and-noise-levels-evgainno3dmsi

http://www.guru3d.com/articles_pages/geforce_gtx_760_inno3d_ichill_herculez_review,6.html

Perhaps too loud at default settings, but those temps... WOW

Looks like the card manufacturers that compete in the US market sit on their laurels compared to Inno3D.

Wish I could get the Inno3D iChill HerculeZ cards (7970) to replace my 7970 and 7950s.

I wonder how much more the models with that cooler cost compared to the stock models?

EDIT:
It looks like they are just using off the shelf Artic Cooling hardware. Now I really want to know how much extra they charge for the iChill.
 
Last edited:

BallaTheFeared

Diamond Member
Nov 15, 2010
8,115
0
71
It looks like the 770 Lightning has better cooling.

iChill is 4C cooler, but it's also 1.2 dB louder, and cooling a weaker chip 760 vs 770...

Maybe they'll release a lightning 760 for you
 

zlejedi

Senior member
Mar 23, 2009
303
0
0
Well it has been double confirmed

Inno3D iChill cooler is just crazy:

http://uk.hardware.info/reviews/452...egment-cooling-and-noise-levels-evgainno3dmsi

http://www.guru3d.com/articles_pages/geforce_gtx_760_inno3d_ichill_herculez_review,6.html

Perhaps too loud at default settings, but those temps... WOW

3 fans like Windforce rev 1.0,
but 5 Al heatpipes, 2x85, 1x 75mm fans vs 3 heatpipes, 3x75mm fans
http://www.computerbase.de/artikel/grafikkarten/2013/nvidia-geforce-gtx-760-im-test/3/

Doesn't look so good compared to MSI - sure it's colder but at big cost on noise.
 

RussianSensation

Elite Member
Sep 5, 2003
19,458
765
126
Please note that the chance of getting a stable 1100+ core clock overclock on 7950 is roughly equivalent chance of getting a 1300-1400 core clock (not set clock, but observed boost clock) 760/770.

This is not true 1350-1400mhz 770 is golden sample territory. This doesn't start for HD7950 until 1225-1275mhz. 1100mhz 7950 is a walk in the park. Go check almost any HD7950 review with solid after-market 7950s or check forums with users who have overclocked their 7950s, specifically Windforce, Sapphire DX, Vapor-X, HIS, MSI TF3 7950s. Now if you want to run your 7950/7970 cards below 1.19V, that's your decision. The chips are safe at 1.256V.
 

96Firebird

Diamond Member
Nov 8, 2010
5,712
316
126
^Thanks for steering the thread off-topic once again.

I like what Gigabyte has been doing with their Windforce coolers, I love their cooler on my ol' GTX 460.
 
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |