6950 2GB Sapphire Toxic OC Review [H]

Page 3 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Madcatatlas

Golden Member
Feb 22, 2010
1,155
0
0
TY for putting it down in writing SlowSpyder, its just such a hard difference for some to see..

Im not going to read that review, for me its totally irrelevant. A 6950 toxic THIS late, is just sad imo. In addition i buy based on price and performance, and i know my 200 $ will get me more performance in a few months than what is being offered by both brands at this pricepoint.


And ofcourse one should be able to discuss competitors at the same pricepoint/performancepoint. The 570 however is a card riddled with faults, including bad drivers and bad pcb. People who buy that card, are taking a that much bigger risk. Id definetly opt for a 2gb 6950 or 6970 if i had the choice of those 3 cards.
 

GoodRevrnd

Diamond Member
Dec 27, 2001
6,801
581
126
I snagged one of these a few weeks ago. I have it seemingly stable at 950 / 1400, stock voltage. I haven't tweaked to max yet--975 crashed furmark and I haven't bothered past 1400 on the memory. I also really haven't put a satisfactory number of gaming hours on it at these clocks yet to be sure of stability, though. It might be fun to try and get this thing past 1000, but I don't want to stress it too hard. Temps peak out absolute max at 88C. What do you think I could take the voltage to with no real ill effects?


Also, other than using slightly old drivers, maybe this review gives a fairer comparison?
http://www.overclockersclub.com/reviews/sapphire_hd6950_toxic_2gb/4.htm
 
Last edited:

happy medium

Lifer
Jun 8, 2003
14,387
480
126
I dont recommend using furmark. Play a few games with the setting maxed.
88c is about as hot as I would run it, but it might run cooler with games.

I think 1.3v is the max you should go and thats only if temps are under 90c ,but really thats pushing it.

I have seen the memory go to as high as 1500, but run benchmarks and see if you start to lose performance after 1400. I think the card has error correction memory which wont artifact or crash but will run slower if you go to high.
 
Last edited:

GoodRevrnd

Diamond Member
Dec 27, 2001
6,801
581
126
Shouldn't use furmark because it will cause me to tend to push it further than it should or because it puts unrealistic stress on the card so I won't go as far as I could? It's just difficult to not use synthetic benchmarks for OCing until you get to the fine tuning stage.

And ya, 88 was a furmark temp. It didn't pass 75 in DoW2 or 3dmark.
 

happy medium

Lifer
Jun 8, 2003
14,387
480
126
Shouldn't use furmark because it will cause me to tend to push it further than it should or because it puts unrealistic stress on the card so I won't go as far as I could? It's just difficult to not use synthetic benchmarks for OCing until you get to the fine tuning stage.

And ya, 88 was a furmark temp. It didn't pass 75 in DoW2 or 3dmark.

Both reasons :)

I used the Crysis demo and let it run the benchmark demo in a loop many times. I also used Metro 1333. As long as the cards are at 100% when running the games its fine.
 

BrentJ

Member
Jul 17, 2003
135
6
76
www.hardocp.com
It's just a bad review, [H] are unable to review anything without some sort of bias. It's a shame really as they used to be a gold standard for video card reviews.

Please email me and explain to me how this review is "just a bad review" please. We welcome constructive criticism, but your post is lacking any detail. We put a lot of time and energy into this evaluation to get it right, from discovering the performance issue, to working with Sapphire to get it fixed, to testing, and reporting as many performance combinations that the card is capable of so you know what to expect with it.
 

WT

Diamond Member
Sep 21, 2000
4,816
59
91
Ordered it and it looks like I got the last one in stock, as I refreshed my wish list a few seconds after buying it and it said 'Out of Stock' yet again.
After what seems like a lifetime of Nvidia, I've decided to give AMD a try. I'll avoid the mudslinging and offer a thank you to Brent @ H-OCP for his review and work on cleaning up the BIOS issue for owners of this card.
 

Grooveriding

Diamond Member
Dec 25, 2008
9,110
1,260
126
Ordered it and it looks like I got the last one in stock, as I refreshed my wish list a few seconds after buying it and it said 'Out of Stock' yet again.
After what seems like a lifetime of Nvidia, I've decided to give AMD a try. I'll avoid the mudslinging and offer a thank you to Brent @ H-OCP for his review and work on cleaning up the BIOS issue for owners of this card.

Let us know how it overclocks!
 

badb0y

Diamond Member
Feb 22, 2010
4,015
30
91
It's just a bad review, [H] are unable to review anything without some sort of bias. It's a shame really as they used to be a gold standard for video card reviews.
If it's does not support my ideas it must be biased.

People quickly forget how [H] crapped on AMD for their Crossfire drivers when they evaluated multi-GPU solutions with the GTX 480s and HD 5870s.
 

GoodRevrnd

Diamond Member
Dec 27, 2001
6,801
581
126
Alright, I've got "short term" stability at 970 / 1380 @ 1.268v, 87C load, 25.5C ambient. I did not attempt voltage above 1.275. "Short term" = 10 Crysis loops.

BUT, what would cause my FPS to DROP when increasing the memory clock? (34FPS @ 1300 vs 28FPS @ 1380)
 
Last edited:

happy medium

Lifer
Jun 8, 2003
14,387
480
126
Alright, I've got "short term" stability at 970 / 1380 @ 1.268v, 87C load, 25.5C ambient...

BUT, what would cause my FPS to DROP when increasing the memory clock? (34FPS @ 1300 vs 28FPS @ 1380)

Error correction in the memory
try 1280 and go up 10hz till you get a drop in performance, then back it off 15 mhz and you have your sweetspot.
 

RussianSensation

Elite Member
Sep 5, 2003
19,458
765
126
Please email me and explain to me how this review is "just a bad review" please. We welcome constructive criticism, but your post is lacking any detail. We put a lot of time and energy into this evaluation to get it right, from discovering the performance issue, to working with Sapphire to get it fixed, to testing, and reporting as many performance combinations that the card is capable of so you know what to expect with it.

Brent, he is prob. just upset that a much cheaper AMD card is whopping NV's flagship. :D
 

GoodRevrnd

Diamond Member
Dec 27, 2001
6,801
581
126
I suspected something like that was the case.

Alright, "final" clocks of 970 / 1340 @ 1.268v. I guess my memory is a dud as it seems most people are able to push 1400, with a few even approaching 1500 (which I find very hard to believe).
 

chihlidog

Senior member
Apr 12, 2011
884
1
81
The lack of stock on these is frustrating. Seems the game deal has expired, which was a big part of why I wanted it, it was a great deal with both games. A few people on [H] mentioned seeing Deus Ex still available with it, am I missing something?
 

Dribble

Platinum Member
Aug 9, 2005
2,076
611
136
If it's does not support my ideas it must be biased.

People quickly forget how [H] crapped on AMD for their Crossfire drivers when they evaluated multi-GPU solutions with the GTX 480s and HD 5870s.

I do remember that, but then I more recently remember their somewhat irrational hatred of crytek just because a game was hard to install. It's a two edged sword - that arrogance that allowed them to stand up to nvidia or ati or whoever was great when used for the right reasons, but these days it has just become an arrogance to say or do whatever they like. That's a shame.

[H] reviewers have become fanboys with a bigger audience, and if you don't like it well you know where the door is. This doesn't sit well with trying to produce professional review of anything. In this case the conclusion should have either been based on all the cards been o/c or none of them. As they picked o/c they should have also made more of the fact there are no guarantee's - you might get lucky and it'll unlock and o/c to 6970 levels, you might not.

Brent - I take it you don't really want my opinion because if I'd posted that on [H] you'd have banned me or at least told me where to go. Not even sure why you are here as you don't even like anandtech as far as I know - I never see you linking to anandtech articles on your site. If you want to talk further feel free to PM me.
 

happy medium

Lifer
Jun 8, 2003
14,387
480
126
Brent - I take it you don't really want my opinion because if I'd posted that on [H] you'd have banned me or at least told me where to go.

In all fairness, I give Kyle and Brent the business on a regular basis on alot of issues with there reviews and they never banned me or even threatened me. In fact I did the same about this article on there forum,in the 6950 Toxic thread.

I don't think he wants to debate anyone on this forum, in public is all, that could be dangerous.

WHo knows mabe they will listen and see the light?

But one thing for sure, I do get attacked by rabbid fanboys who back up everthing Kyle and Brent do. :)
 
Last edited:

WT

Diamond Member
Sep 21, 2000
4,816
59
91
A few people on [H] mentioned seeing Deus Ex still available with it, am I missing something?

Chihli, when I bought mine yesterday early AM, you had to add the free game to the card as a 'combo' purchase. Kinda sneaky of Newegg to not show that up front, but that was what was happening. It shows up as a $49 price deduction for the $49 game.
 

notty22

Diamond Member
Jan 1, 2010
3,375
0
0
I do remember that, but then I more recently remember their somewhat irrational hatred of crytek just because a game was hard to install. It's a two edged sword - that arrogance that allowed them to stand up to nvidia or ati or whoever was great when used for the right reasons, but these days it has just become an arrogance to say or do whatever they like. That's a shame.

[H] reviewers have become fanboys with a bigger audience, and if you don't like it well you know where the door is. This doesn't sit well with trying to produce professional review of anything. In this case the conclusion should have either been based on all the cards been o/c or none of them. As they picked o/c they should have also made more of the fact there are no guarantee's - you might get lucky and it'll unlock and o/c to 6970 levels, you might not.

Brent - I take it you don't really want my opinion because if I'd posted that on [H] you'd have banned me or at least told me where to go. Not even sure why you are here as you don't even like anandtech as far as I know - I never see you linking to anandtech articles on your site. If you want to talk further feel free to PM me.
Those are some fair points. Two reviews, both encountered difficulties.
With both they found fixes and implemented them, to get the reviews done.
One they pat themselves on the back (6950 Toxic), and with the Free DX11 Patch, rant and brand it with the Gold Fail Award. LOL

Other review sites have failed to identify this performance problem, while most do not even compare the SAPPHIRE TOXIC HD6950 to a stock Radeon HD 6950. By finding this problem and notifying SAPPHIRE of the flaw that this cards BIOS has, we're directly helping the community. During our testing time SAPPHIRE worked up a new BIOS fix, so that consumers do not have to follow the solution below that we used prior to receiving the BIOS fix. Unfortunately, anyone that owns this card with the current on-shelf BIOS will have to manually flash their BIOS to the new version to solve this problem. SAPPHIRE will be releasing this BIOS publically so you can flash your video card, and won't have to resort to the method we used to fix the issue below, which is an alternative way of getting around the issue.
 

3DVagabond

Lifer
Aug 10, 2009
11,951
204
106
I read another review, don't remember which site, but they too said to move the power slider up to stop throttling of the card. They said it sucked, but didn't say they had contacted anyone for a fix or not. They said they had no idea whether there would be a bios update or not. For them the fix was the power slider.

It makes you wonder what kind of testing cards go through? This should have been noticed in the prototype testing. Unless it's something to do with the whql drivers power limiting, and they only test with in house drivers? Don't know, but it's pretty obvious and shouldn't require a comparison with any other card to find. I believe the other site used GPU-Z and saw the throttling occurring.
 

Madcatatlas

Golden Member
Feb 22, 2010
1,155
0
0
I do remember that, but then I more recently remember their somewhat irrational hatred of crytek just because a game was hard to install. It's a two edged sword - that arrogance that allowed them to stand up to nvidia or ati or whoever was great when used for the right reasons, but these days it has just become an arrogance to say or do whatever they like. That's a shame.

[H] reviewers have become fanboys with a bigger audience, and if you don't like it well you know where the door is. This doesn't sit well with trying to produce professional review of anything. In this case the conclusion should have either been based on all the cards been o/c or none of them. As they picked o/c they should have also made more of the fact there are no guarantee's - you might get lucky and it'll unlock and o/c to 6970 levels, you might not.

Brent - I take it you don't really want my opinion because if I'd posted that on [H] you'd have banned me or at least told me where to go. Not even sure why you are here as you don't even like anandtech as far as I know - I never see you linking to anandtech articles on your site. If you want to talk further feel free to PM me.



Hardocp reviews are great examples of how you should do testing, by first explaining your methodology and eventual sidesteps from that methodology if the hardware doesnt leave any other option etc and then testing that hardware with different game engines and doing apples to apples comparisons. Another thing that makes the [H] stand out, is their testing at higher resolutions with higher end hardware.

On several occasions Hardocp even contacts Nvidia or AMD and asks them for a fix to a problem on the fly or before they post the review, and THIS is also noted in the methodology bit, as it should. Id go as far as to say that noone (with the exception of Anands own and Ryan Smiths reviews) explains their methodology better than Hardocp.

I cant stand the dribble that is a 30 page long testing of games at 1200x800 or whatever with a 580 or a 6970 which alienbabeltech does for example. We really need to know how a Radeon 6990 does at that resolution in Farcry or better yet, Hawx (lol..).


The only relevance i see in this review, seeing as we are closing fast on the launch of the 7xxx series, is that while there are some 6970s in stock, several stores are showing 0 in stock, some dont even have a single Sapphire 6970 even SHOWING as a product.
This card at these clocks and speeds, is essentially a 6970 for the price of a 6950 or abit higher if im not too wrong. The card is also made for overclocking it seems and has some added strenghts like a better cooler and probably also some higher quality components.

Still, not enough to make me change graphic cards. Ill hold on till the 7xxx series.
 

BladeVenom

Lifer
Jun 2, 2005
13,365
16
0
Hardocp reviews are great examples of how you should do testing, by first explaining your methodology and eventual sidesteps from that methodology if the hardware doesnt leave any other option etc and then testing that hardware with different game engines and doing apples to apples comparisons. Another thing that makes the [H] stand out, is their testing at higher resolutions with higher end hardware.

+1, here and Hardocp are where I get most of my reviews from.