And people moan about online reviews...

N

Neeyik

Homo ergaster
Veteran
I had the "pleasure" of reading the May 2004 issue of PCW UK yesterday at work. It had a 10 graphics card shoot-out, mostly FX 5950 Ultras against 9800 XTs. To analyse the cards in detail, they tested them using 3DMark03 with and without AA, UT2003 with and without AA plus PCMark04...and that's it. The entire performance of each product was judged purely on what 3DMark score each produced at 1024x768 (fps for UT2003, of course).

Sales of printed media may have taken a dint with the rise of online review but PCW still shifts a great deal of magazines; you'd think that they would attempt to come up with some better - especially since the shoot-out was a key feature of that issue.

And to think I once bought a Creative Voodoo2 12MB based on their review!
 
I swore off of Maximum PC last year after they proclaimed the FX 5950s as winner for their high-end rig contest based on 3DMark03 and UT2003 scores. No mention of the ongoing Detonator cheating, which was very well discussed by time of publication, or of the questionable fp performance in the nVidia chips. Flash forward 8-9 months and I hear about how they redressed the situation in the very next issue and how I should give them another chance, so I bought their May issue for its 6800U coverage. On page 32 is a sidebar about on R420 that reads:

We expect R420 to be an updated 0.13-micron version of the R360 core that powered the Radeon 9800 XT. . . .While we expect to see some minor new features, we'd be surprised if ATI incorporates any major changes other than the move to a smaller process. . . .We've also received reliable reports that R420 will not support Pixel Shader 3.0 and Vertex Shader 3.0, and that it will be limited to 24-bit floating-point color throughout its pipeline (the NV40 supports 32-bit floating-point). While R420 will still run Pixel and Vertex Shader 2.0 code, without 3.0 support for code loops and branches, developers will have to break long Pixel Shader 3.0 shader programs into small, easily digestible chunks that will run on the R420. This will absorb clock cycles and could seriously affect performance in games that use more sophisticated programmable pixel shaders.
Click to expand...

Sigh. Definitely the last issue of that rag I ever buy.
 
John Reynolds said:
We expect R420 to be an updated 0.13-micron version of the R360 core that powered the Radeon 9800 XT. . . .While we expect to see some minor new features, we'd be surprised if ATI incorporates any major changes other than the move to a smaller process. . . .We've also received reliable reports that R420 will not support Pixel Shader 3.0 and Vertex Shader 3.0, and that it will be limited to 24-bit floating-point color throughout its pipeline (the NV40 supports 32-bit floating-point). While R420 will still run Pixel and Vertex Shader 2.0 code, without 3.0 support for code loops and branches, developers will have to break long Pixel Shader 3.0 shader programs into small, easily digestible chunks that will run on the R420. This will absorb clock cycles and could seriously affect performance in games that use more sophisticated programmable pixel shaders.
Click to expand...

Sigh. Definitely the last issue of that rag I ever buy.
Click to expand...
What exactly about that paragraph made you swear off reading MaximumPC? Given they only had a paragraph to describe rumors it doesn't seem bad to me. Am I missing something?
 
3dcgi said:
John Reynolds said:
We expect R420 to be an updated 0.13-micron version of the R360 core that powered the Radeon 9800 XT. . . .While we expect to see some minor new features, we'd be surprised if ATI incorporates any major changes other than the move to a smaller process. . . .We've also received reliable reports that R420 will not support Pixel Shader 3.0 and Vertex Shader 3.0, and that it will be limited to 24-bit floating-point color throughout its pipeline (the NV40 supports 32-bit floating-point). While R420 will still run Pixel and Vertex Shader 2.0 code, without 3.0 support for code loops and branches, developers will have to break long Pixel Shader 3.0 shader programs into small, easily digestible chunks that will run on the R420. This will absorb clock cycles and could seriously affect performance in games that use more sophisticated programmable pixel shaders.
Click to expand...

Sigh. Definitely the last issue of that rag I ever buy.
Click to expand...
What exactly about that paragraph made you swear off reading MaximumPC? Given they only had a paragraph to describe rumors it doesn't seem bad to me. Am I missing something?
Click to expand...

After pages praising the 6800U I thought the overall tone of the sidebar was unprofessional and unfair. The X800 chips aren't just .13u refreshes of the R360 as it erroneously implies, and no game is going to ship with shaders longer than the instruction lengths supported by them. Sweeney mentioned that UE3's current shaders are 50-150 instructions, and if you consider the first game using that engine won't ship until 2006. . . . I don't know, just seemed like very one-sided FUD and I think a print mag with their circulation should be more careful in what they print.
 
John Reynolds said:
After pages praising the 6800U I thought the overall tone of the sidebar was unprofessional and unfair. The X800 chips aren't just .13u refreshes of the R360 as it erroneously implies, and no game is going to ship with shaders longer than the instruction lengths supported by them. Sweeney mentioned that UE3's current shaders are 50-150 instructions, and if you consider the first game using that engine won't ship until 2006. . . . I don't know, just seemed like very one-sided FUD and I think a print mag with their circulation should be more careful in what they print.
Click to expand...

You mean 512instructions isn't a small easily digestable chunk?
 
John Reynolds said:
I swore off of Maximum PC last year after they proclaimed the FX 5950s as winner for their high-end rig contest based on 3DMark03 and UT2003 scores.
Click to expand...
While they don't have a variety of 3D games and apps in their benchmarks I do have to give their graphics card reviewers credit for one thing. They eventually went back and named the 9800 the best available card.
 
Wow, they still have magazines? I gave up on magazines back in 1999, when I could read the same information months before it appeared in print.
 
Unfortunately, most magazines as far as gfx cards are concern still live in 2001 and bench accordingly. I should know, I work in one :p
 
Funnily enough, I was going to start a topic like this just the other day:

Last week, I was reading through the latest PC Pro magazine, where they had a page crowing from the rooftops how they had 'toughened up' their graphics card test to help buyers make a more informed choice.

What was their new suite of comprehensive benchmarks?

UT2004 (to test DirectX 9 performance, no less :| )
Halo's timedemo (that's right, the timedemo that isn't even reflective of in-game performance)

So, two whole games... Incredibly thorough. :rolleyes:
 
PC Pro is actually a pretty good magazine (I have a subscription to them) but they remain pretty clueless when it comes to graphics cards.

When all the cheating/optimisation stuff came out about the NV cards they equally blamed both ATI and NV for cheating. Ultimately, the reviews are written by people much less knowledgable than most people on this board - they really need someone with more specialist knowledge about the subject. Other than that, it's a pretty decent mag.
 
Neeyik said:
they tested them using 3DMark03 with and without AA, UT2003 with and without AA plus PCMark04...and that's it
Click to expand...
Please tell me you're kidding me about the inclusion of PCMark04 :oops:
 
Diplo said:
Neeyik said:
they tested them using 3DMark03 with and without AA, UT2003 with and without AA plus PCMark04...and that's it
Click to expand...
Please tell me you're kidding me about the inclusion of PCMark04 :oops:
Click to expand...
Seriously, yes - that's what was used. I know that printed media uses a lot of paper coverage for advertising and that most articles need to be fairly short but they sure as hell didn't spend much time testing the cards. You get all 10 day finished in one day!
 
Mariner said:
PC Pro is actually a pretty good magazine (I have a subscription to them) but they remain pretty clueless when it comes to graphics cards.
Click to expand...

Pretty much hits the nail on the head - I like the magazine, don't get me wrong, but sometimes I can't help but weep when I see their video card reviews.
 
You must log in or register to reply here.

Similar threads

R
Reverend at The Pulpit #5
2
Replies
22
Views
6K
Joe DeFuria
J
Back
Top