SemiAccurate Forums  

 
Go Back   SemiAccurate Forums > Main Category > GPUs

GPUs Talk about graphics, cards, chips and technologies

Reply
 
Thread Tools Display Modes
  #641  
Old 03-24-2012, 03:06 PM
Zechsy Zechsy is offline
word
 
Join Date: May 2011
Posts: 20
Zechsy is on a distinguished road
Default

Quote:
Originally Posted by James View Post
Lots of stuff.
Maybe it's just me but comparing a custom-edition ("Lightning") design to reference designs seems rather unreasonable. The "Lightning" Edition used in the hardwareheaven review is clocked at 1265/1505MHz, 100MHz higher than the highest one on the list below.

Here's a roundup of overclock results of reference HD7970s.



With that said. Here are some 5760x1080p reviews, overclocked and non-overclocked from other, imo, more reasonable reviewers.

OverclockersClub.com

NV (Stock) wins: 6/6. NV (OCvsOC) wins: 5/6, 1 draw.

HardOCP.com

NV(Stock) wins: 3/5. The other 2 were lost by 2-3FPS.

pcper.com

NV(Stock) wins: 3/4.

eteknix.com

NV(Stock) wins: 4/6.

Quote:
Originally Posted by vain View Post
More stuff.
Hmm, time to dress up like Sherlock and get to the bottom of this. Will you be my Watson?

Last edited by Zechsy; 03-24-2012 at 03:25 PM. Reason: Other stuff.
Reply With Quote
  #642  
Old 03-24-2012, 03:12 PM
CAT-THE-FIFTH CAT-THE-FIFTH is offline
itanic
 
Join Date: Jun 2011
Posts: 148
CAT-THE-FIFTH is an unknown quantity at this point
Default

I am going to ask a noob question here.

So each GTX680 card does not turbo upto a preset maximum,but the maximum is sample dependent?? Hence,stock performance will vary between cards upto a degree??

If,so is there any website which has a table like in post 636,which shows the range of stock maximum turbo speeds??
Reply With Quote
  #643  
Old 03-24-2012, 03:30 PM
James James is offline
2^11
 
Join Date: Jul 2009
Posts: 3,068
James will become famous soon enough
Default

Quote:
Originally Posted by Zechsy View Post
Maybe it's just me but comparing a custom-edition ("Lightning") design to reference designs seems rather unreasonable. The "Lightning" Edition used in the hardwareheaven review is clocked at 1265/1505MHz, 100MHz higher than the highest one on the list below.
I think it's unreasonable to allow Nvidia's card to overclock themselves with no way of disabling it. That's what the vast majority of reviews have shown - a highly overclocked cherry picked card vs stock 7970's.

Btw most of those OC's on the 7970's were without any voltage increases.

Last edited by James; 03-24-2012 at 03:39 PM.
Reply With Quote
  #644  
Old 03-24-2012, 03:33 PM
James James is offline
2^11
 
Join Date: Jul 2009
Posts: 3,068
James will become famous soon enough
Default

Quote:
Originally Posted by CAT-THE-FIFTH View Post
I am going to ask a noob question here.

So each GTX680 card does not turbo upto a preset maximum,but the maximum is sample dependent?? Hence,stock performance will vary between cards upto a degree??

If,so is there any website which has a table like in post 636,which shows the range of stock maximum turbo speeds??
Stock (turbo) and overclocking performance will vary. That's the thing - these 1300 MHz cards could easily be 1050 MHz max...or less. Nobody should be happy about this. If I were an Nvidia fan I wouldn't be remotely interested in buying a card that could have such variability in performance.

Most reviewers are benching in open test benches as well. How much does heat affect the OC?

And other thing - the heat will rise as you continue to play, lowering overall fps scores the longer you are playing. There is no way that is showing up in these benchmarks on open test benches, with benchmarks lasting a couple of minutes each.

Last edited by James; 03-24-2012 at 03:42 PM.
Reply With Quote
  #645  
Old 03-24-2012, 03:42 PM
CAT-THE-FIFTH CAT-THE-FIFTH is offline
itanic
 
Join Date: Jun 2011
Posts: 148
CAT-THE-FIFTH is an unknown quantity at this point
Default

Quote:
Originally Posted by James View Post
Stock (turbo) and overclocking performance will vary. That's the thing - these 1300 MHz cards could easily be 1050 MHz max...or less. Nobody should be happy about this. If I were an Nvidia fan I wouldn't be remotely interested in buying a card that could have such variability in performance.

Most reviewers are benching in open test benches as well. How much does heat affect the OC?
So this is not like CPU Turbo which has an upper limit AFAIK??

If so,how can you get reliable figures for performance as it would vary from batch to batch??
Reply With Quote
  #646  
Old 03-24-2012, 03:50 PM
James James is offline
2^11
 
Join Date: Jul 2009
Posts: 3,068
James will become famous soon enough
Default

Quote:
Originally Posted by CAT-THE-FIFTH View Post
If so,how can you get reliable figures for performance as it would vary from batch to batch??
You can't. Nvidia doesn't care, they've already got their fawning tech press reviews with their cherry picked cards.
Reply With Quote
  #647  
Old 03-24-2012, 07:11 PM
SkyQuake SkyQuake is offline
word
 
Join Date: Feb 2012
Posts: 22
SkyQuake is on a distinguished road
Default

James
I think its you who looks trough tinted glas I dont care about red or green-team just tech and horsepower.

I see you try to proove your point with percentages and the graphs. They both win and loose some - i did not state otherwise. I just want to understand why people was so certain 2 GB and small bus wuld = bad performance @ high resolutions. I diddent know - so i figured they were right. Sounded plausible. But in any case i dont find the GTX 680 to fail so miserably as people said it wuld due to its apperent hardware-specifications (bus/ram).
I still dont get how it can be so competitive. I wuld have belived it to rock in 1080, but not at higher res. At least that was all the negative hype up untill its release.
Reply With Quote
  #648  
Old 03-24-2012, 07:51 PM
kac kac is offline
itanic
 
Join Date: Feb 2010
Posts: 168
kac is on a distinguished road
Default

Quote:
Originally Posted by SkyQuake View Post
James
I think its you who looks trough tinted glas I dont care about red or green-team just tech and horsepower.

I see you try to proove your point with percentages and the graphs. They both win and loose some - i did not state otherwise. I just want to understand why people was so certain 2 GB and small bus wuld = bad performance @ high resolutions. I diddent know - so i figured they were right. Sounded plausible. But in any case i dont find the GTX 680 to fail so miserably as people said it wuld due to its apperent hardware-specifications (bus/ram).
I still dont get how it can be so competitive. I wuld have belived it to rock in 1080, but not at higher res. At least that was all the negative hype up untill its release.
That's not what James is getting at. This is definately not a red vs green thing. This is the problem. If you bought a 680 and paid $500 and I bought a 680 and paid $500 we would experience different performance given the same testing environment. One of us, you or me could be recieving less for the money we spent.

This is different than the way Intel implements it's TurboCore features. If you bought a 920 and I did as well and we (given the same testing environment) by and large we would experience the same boost effect within a couple of percentage points. The clock rate is locked in based on resources used. If both you and I reached Step 5 we would both be at 3.2 GHz. All Intel CPU's at a given clock rate are certified to hit a top clockrate.

From what I can gather (but someone correct me if I'm wrong) the amount of boost is directly tied to the quality of the core and thus temp level that is experienced when playing whatever game. Intel comparitively guarantees clock rate increases by step at a given load all things being equal. Nvidia isn't guaranteeing a top clock. It could be anything above 1050 which is what's creating the problem. Intel releases the clock rate steps for all models and supposedly all models should be able to reach the same top clock rate all things being equal.
Reply With Quote
  #649  
Old 03-24-2012, 08:13 PM
integrated integrated is offline
2^10
 
Join Date: Dec 2009
Posts: 1,477
integrated will become famous soon enough
Default

Quote:
Originally Posted by James View Post
Stock (turbo) and overclocking performance will vary. That's the thing - these 1300 MHz cards could easily be 1050 MHz max...or less. Nobody should be happy about this. If I were an Nvidia fan I wouldn't be remotely interested in buying a card that could have such variability in performance.

Most reviewers are benching in open test benches as well. How much does heat affect the OC?

And other thing - the heat will rise as you continue to play, lowering overall fps scores the longer you are playing. There is no way that is showing up in these benchmarks on open test benches, with benchmarks lasting a couple of minutes each.

If you were a fan, you wouldn't question the win
Reply With Quote
  #650  
Old 03-24-2012, 08:19 PM
integrated integrated is offline
2^10
 
Join Date: Dec 2009
Posts: 1,477
integrated will become famous soon enough
Default

Quote:
Originally Posted by Drunkenmaster View Post
It really depends, I don't think we'll see an "average" performance till we've seen thousands of cards sold and used for a month +. Even then unfortunately we'll see mostly "its so much faster than my last card" rather than, its not as fast as the review cards. I said earlier, with a 680gtx review you can't be certain you'll get even within 10% of those scores, potentially even less than that, with a 7970 you know what you're getting. It's pretty odd, its the same as if every review just did max clocks the card could do as that cards "normal" clock, and every review was different.

...snip...

It is a lot like that. Nobody knows what a stock 680 performs like, nobody!

Somebody in this thread quoted Nvidia as stating that the average boost amounts to 1056(?)Mhz, meaning that every review card is a highly exceptional "golden sample" carefully binned for the launch.

10-20% OC has completely tipped the balance in these reviews, and the reviewers are helpless to avoid it (although they could have made it clearer to the reader!)


It is exactly the same situation as a hypothetical Pitcairn launch in which all the cards were OC'd to max stable clocks, but not labelled as such on the graphs.

Everyone would see Pitcairn destroy the GTX580, then go out and buy one that couldn't match a 570...
Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off


Forum Jump


All times are GMT -5. The time now is 05:23 PM.


Powered by vBulletin® Version 3.8.7
Copyright ©2000 - 2014, vBulletin Solutions, Inc.
SemiAccurate is a division of Stone Arch Networking Services, Inc. Copyright 2009 Stone Arch Networking Services, Inc, all rights reserved.