BRIGHT SIDE OF NEWS About | Advertise | Contact BSN USER Login
| Register 		SUBSCRIBE Newsletter | RSS Feeds
Thursday, March 18, 2010
Email this to a friend.
Your friend's e-mail:
Your Name:
Your e-mail:
Message subject:
Comments
by: Kyocera on 11/20/2009
borrowed from Net:
"I just had a really loud laugh; ATI is peeling the wallets of their European fans,; HD 5970 was sold out in the matter of hours for 850-900 dollars.
An article on Bright-side-of-news entitled "Are all AMD fans idiots" just came to my mind."

"Yup, it sucks if you are unable to resolve the quality insurance across the board and partners and have to downscale the flagship product as ATI did with 5970. :)) :)) :)) Same as AMD processors, they are "almost as fast" as Intel's but much "greener" (do not go over 25W)."

my comment? :
maybe not AMD fans are idiots, but the European ones certainly proved to be heavy ones.
For $600+ it should by: Michael A. McKenney on 11/20/2009
$600+, it should stomp every other card along with my credit card.
@Scores by: Sean Kalinich on 11/20/2009
Which Scores are you talking about?

If you are talking about gaming we do full in game testing not pre-rendered synthetics.

If you are talking about Stalker then that is due to many sites not running the proper Image Quality or any form of AA. [to be honest many sites do not run games with AA on]
Scores by: Anonymous on 11/20/2009
how come ur results r lower than other review sites on this 5970 graphics card...

does it have to do with your motherboard?
Flawed benchmarks by: Theo Valich on 11/20/2009
As far as Flawed Benchmarks go,

If anyone reading this is able to program a better DirectCompute benchmark, please let us know.

Put your money where your mouth is. We are now testing out a new GPGPU benchmark that is based on real-world engine that generates hundreds of millions of dollars in revenue.

Oddly enough, the developer of that benchmark app told us the same thing as far as scaling goes. Now, are we going to call that engine flawed?

Bottom line is - AMD's HD5970 didn't run DirectCompute benchmark by using two GPUs, since HD5870 returns a bit better score [I got 33700]. The problem with the calculation is that it targets those 320 cores that work at 725 / 850 MHz and that goes against nVidia's 240 shaders at 1.5 GHz+.

This is not fishy or looking weird, its just that the load made by the benchmark is worst possible one for ATI architecture, and that wasn't deliberate, that was an FFT. The benchmark we're evaluating features even worse case scenarios for GPUs as such.

Ed.
Graphs by: Sean Kalinich on 11/19/2009
The larger graph [the one with the black background and colored bars] shows DX10 performance. again the average FPS of each, minimum and max frame rates are not as important as average.

The one that looks like a stand alone table shows the HD5970 average FPS performance in DX11.

There is a note below the two tables that shows the differences between the two results.
@ Didn't mention by: Sean Kalinich on 11/19/2009
9.11 does not support the HD970.
take a look on AMD's website, you will find that out. AMD offers only the 8.663.1_Beta5_Hemlock_VistaWin7_Nov11.exe for the HD5970.

https://a248.e.akamai.net/f/674/9206/0/www2.ati.com/drivers/hotfix/8.663.1_Beta5_Hemlock_VistaWin7_Nov11.exe

Strange, I have not heard from anyone that the DC bench is flawed.. care to show where you get that from?

Also I have a feeling you are talking about the Stalker bench. Are you reading Average FPS or max.

We reported average FPS as shown in the review not MAX. If you are really getting better performance with a single 5870 please post the link to it.

Or is it this one already?
http://www.evga.com/forums/tm.aspx?high=&m=29043&mpage=1#29741

if you look at the screen shot that was posted there in this post

http://www.evga.com/forums/fb.ashx?m=29855

you can see that my results are in fact better than your single card run

You have

Day -
Min 35.7 Average 47.7 Max 67.4
Night -
Min 29.6 Average 42.2 Max 61.4
Rain -
Min 31.0 Average 44.6 Max 62.5
SunShafts -
Min 22.5 Average 30.4 Max 45.1

I have

Day -
Min 31.2 Average 63.6 Max 82.2
Night -
Min 35.8 Average 55.2 Max 76.6
Rain -
Min 32.4 Average 57.8 Max 78.2
SunShafts -
Min 26.1 Average 41.3 Max 61.4

I am not sure how you equate your results to better than what I got out of the HD5970
didn't mention by: Anonymous on 11/19/2009
Sean, you didn't mention that the DirectCompute benchmark was not applicable as a viable benchmark. You might as well have not even included it since it has already been determined amongst various places on the NET to be "FLAWED". How you can use a flawed benchmark in your review and not mention what is wrong with even attempting to compare a DX10 GPU to a DX11 GPU in this benchmark is wreckless reviewing.

It is odd how I get better results when running a single GPU 5870 than you do in some of your tests when running the 5970 (dual-GPU.

You do realize that you should have used the 9.11 WHQL driver in this review instead of the 8.663.1 drivers, correct?
by: Kyocera on 11/18/2009
Thanks for the explanation Sean :) :) :)

(I enjoyed your test and article! from all aspects; the hardware side and the large view approach.)

P.S:
I must admit I was almost roaring with laughter when reading the gliding accident; well, till the second part and till I saw who was hurt.
If Ivica is half as cute as he is nice, he will have the attention of all the nurses there :)) :)) :))
the only thing to see is how is this going to impact his convalescence time :) :)
As it might go, we extend our wishes for an uncomplicated recovery.
Kyocera
@ bad review by: Sean Kalinich on 11/18/2009
you cannot disable CF on the 5970 (I say that in the article).

As far as your accusation that I changed the name of the furmark.exe you are completely incorrect. I did not alter any benchmarks at all.

Please do not post false accusations.

As far as Direct Compute the benchmark selects the appropriate level. As you can see in the screen shot for the 5900 it recognizes that it is a DX11 card (D3D_FEATURE_LEVEL_11_0) if you select D3D11_CREATE_DEVICE_SINGLETHREADED you disable multi-GPU support I left that unchecked for both tests. you cannot force DX10 DC on an ATI card as it does not support it.
Bad Review by: Anonymous on 11/18/2009
CF was obviously disabled in some of your tests.

Your DirectCompute benchmark was in error seeming that the Nvidia DX10 cards are using D3D_10 ---> Shader Model 4.0, DirectCompute 10 while the ATI DX11 cards are using D3D_11 --> Shader Model 5.0, DirectCompute 11.

Furmark is completely in error because you changed the name of the program from Furmark.exe. That is very malicious. I get double what you spewed out with only a single 5870!!!
ATi Firsts by: Sean Kalinich on 11/18/2009
DX1, DX3, DX5, (there was no DX4), DX6, DX9b, DX10.1, DX11

Also remember that the 6800 claimed it was "Longhorn ready" I have pictures of that on boxes that shipped.

AMD (actually ATi) has always done a good job of supporting DX, they have also always had spotty execution in terms of getting hardware to market (as I mentioned below)
Direct X by: Sean Kalinich on 11/18/2009
Typically AMD has been ahead with DX in terms of initial support. I am not saying better performance but better initial support for the standard. Direct X was co-built by ATi, MS, Cirrus Logic and S3.

"The video card hardware folks LOVED the idea of something that took advantage of their hardware. ATI, Cirrus Logic, and S3 (there were others I'm sure but those are three I remember) all came up to the Microsoft porting lab in building 20 to get their hardware to work with DirectDraw (I lived in building 20 with those guys for almost 2 months). After a huge push we had a beta ready for April - it only worked on the ATI Mach 64 card (with what was then a huge amount of RAM - 4 megabytes) and a Soundblaster card."

http://craig.theeislers.com/2006/02/directx_then_and_now_part_1.php


Where AMD/ATi have lost out has been in getting their product to the market ahead of the game [the 58xx launch is an exception]. AMD/ATi announced DX10 support for the 2xxx series ahead of the announcement of the 8xxx series from NV. Then [as has been their pattern] failed to get the product to market on time. It was also a terrible product, in fact AMD did not have a competing product until their DX10.1 cards came out about a year later.
direct X by: Anonymous on 11/18/2009
As far as I know,
the first hardware support DirectX8 is Geforce3
the best hardware for DX8.1 is Geforce 4

ATi strikes with its raedon 9700 to be the first hardware compliant to DX 9.0b, while nVidia has the NV30 fiasco.

However, going to the lasting DX 9.0c, the first hardware is Geforce 6800 and ATi has the notorious HDR bug even long after their X1000 series debut (which supposedly to support 9.0c but failed)
Only after Ati pulling out X1900 XTX do they have a proper support of DX9.0c, which is only a month or two before 7900GTX, which is already nVidia's 3rd gen card with DX 9.0c support. Well, X1900 series is the best DX9.0c card though, no doubt...

Thus far, there are only 3 times that Ati has the hardware going out first: 9.0b, 10.1 and 11. I am not denying they are doing a great job today, but putiing it as if it is always the case seems quite far fetched.
@Kyocera by: Sean Kalinich on 11/18/2009
Apple bought a considerable amount of the 4800 ASIC stock from AMD to support their own systems. This made the supply of them for the AIBs very low.

Also the 58xx series was short [as we mention in the article] as such it made it difficult for the AIBs to produce any 58xx cards forcing them to push the 57xx cards instead.
5870 by: Sean Kalinich on 11/18/2009
The reason for not using a 5870 was that I was not included in the sampling of that product.
I do not have one in my lab.

As far as DX10 if you read what I wrote they were the first to have both hardware and drivers that supported DX10 features. NV was late to the game with their DX10 drivers if you remember. Yes the 8800GTX was DX10 hardware but that was of little use without drivers that could do anything. Still, the 8800GTX easily outperformed the 2900XT [as it was a terrible card]

Remember that ATi was one of the companies that helped to created DX in the first place. As such they have had a hand in every revision of DX since then [nV joined them later after the 3dfx purchase if I remember right].
by: Kyocera on 11/18/2009
"Naturally, not so great about AIBs that are now openly hostile towards AMD. In conversations we had with their partners, it looks like the olive branch was the tiny fact that most of HD5700 series goes to AIBs, who are focusing on these boards instead of 4800 [stock robbed by Apple] or 5800."

Would you explain better, please???
Please don't just tell lies by: Anonymous on 11/17/2009
I am not an nVidia fan but it is a very clear misstatement that AMD has the first DX10 part. nVidia's 8800GTX is 6 months ahead of ATI's HD 2900 which is not only late but also a disaster itself.

Besides, while the benchmark is good I can't think of a reason why did you not put a 5870 beside to see the scaling directly. Is it due to the fact that they behave very close?
© 2009 - 2010 Bright Side Of News*, All rights reserved.
Top Stories
Most commented
Highlight
  • AMD forms Fusion strategy, Offensive for 2011 set
  • PowerDVD 10 creates 3D movies from 2D ones, MKV included
  • Updated GTX 470 / 480 Specs Leak Out
  • Apple loses Director York to an aneurysm
  • Apple loses Director York to an aneurysm
Recent news
March 20, 2010, 20:00 UTC

Dear Readers,

In order to enable new features for the site, we'll be temporarily offline on Saturday, March 20th 2010 at noon Pacific, 3PM Eastern or 8PM/20:00 GMT/UTC. We should be offline for 15-25min, after which you should be able to see new features.

Thank You for understanding,

The BSN* Team
© 2009 - 2010 Bright Side Of News*, All rights reserved.