Radeon 9600 next week?

The 42.74 drivers have "balanced" renamed to "quality" and "aggressive" renamed to "performance". I saw it on the demo machines that were running at the NVidia booth @ CeBit. The German language version still has the old names (in German, of course ;) )
 
If they'd simply renamed Application to Quality and Aggressive to Performance the confusion would be resolved...

Any reviewer worth spit would be able to show that Nvidia "Quality" in that case was better than ATI's Quality but having slower performance.... that Balanced was a good tradeoff of speed/quality, and performance was butt ugly compared to ATI's performance, but fast.

Nvidia doesn't appear to have any faith in their own technology however... They have to instead attempt to resort to obfuscation to "win" benchmarks in reviews where the reviewer, well, isn't worth spit.

One question I've not seen answered was weather the "Application" mode would use "bilinear AF" if bilinear was selected in the application... if it doesn't and it actually forces trilinear, then why do they call it Application in the first place?
 
mboeller said:
Code: 
UT2003 Flyby: 1024 no AA, no AF / 1024 2xAA, 8xAF / 1280 2xAA, 8xAF

R9600:               92 / 41 / 29
R9600 Pro:          124 / 55 / 38
GF-FX 5600 Ultra:   122 / 67 / 46
GF-FX 5200 Ultra:    82 / 48 / 32
GF-FX 5200:          49 / 26 / 16
Click to expand...

As it seems the c't benchmarks could be a little bit on the low side.

Tomshardware has posted an article about the notebook-version :
P4 3.06GHz + i845 + 512MB DDR ( =Shuttle-PC ) + MR9600 @350MHz + 300MHz DDR-SDRAM

3DMark03 : 3144
3DMark01 : 10720

UT Asbestos Flyby @ 1280x1024x32
without AA+AF : 148.5
4xAA : 82,87
4xAA + 8xAF : 57.58

Link : http://www.tomshardware.com/mobile/20030324/ati_nvidia-03.html


IMHO, nice. Anyone knows (has an link?) how the 3DMark01 score compares to desktop-chips ( with P4 @ 3.06GHz)
 
Probably should say this, but there is a Sapphire board doing the rounds. I've got it last night and I've had some problems with it, in that I thought it was dead. Anyway, I removed the heatsink from the chip and saw that arctic silver saw smooshed all over the flipchip package - AFAIK arctic silver is conductive, and I thought this wasn't a very good thing!

Anyway, I cleaned up the package as much as possible but it still failed to work. Changed the motherboard and it did work (odd, seeing as the Intel board I use is usually bullet proof in terms of compatibility). Anyway, I ran a few tests and including Humus's GL_reme render order - it seemed as though HierZ wasn't working since its efficiency was about the same as 9500 and far away from 9500 PRO. The resistors round the flipchip package do seem to turn things on or off from other boards, so I'm wondering if all the arctick silver goo all over it has disabled shorted it and disabled HierZ.

I'd take any benchmarks seen from this Sapphire board (which has been a number of places already) with a pinch of salt right now.

And, no, its not going to be next week (officially).
 
BenSkywalker said:
AFAIK arctic silver is conductive, and I thought this wasn't a very good thing!
Click to expand...
It definitely is conductive, not good. Did you try using some rubbing alcohol to clean it off?
Click to expand...
I have seen conflicting reports on the stuff. I know at least one tube I bought was specifically marked 'non conductive'.

I don't quite believe that. I had an Athlon that got somewhat spooged by the stuff and refused to boot, after excessive cleaning with alcohol it worked - but the multiplier had changed.

I suspect it has some electical effect - probably a moderate resistance, not enough to short lines but enough to act as a pull-up/down resistor.
 
I have seen conflicting reports on the stuff. I know at least one tube I bought was specifically marked 'non conductive'.
Click to expand...

Arctic Silver marked non conductive? You sure it wasn't Arctic Alumina? Artic Silver isn't as conductive as some of the other compounds, but it definitely is conductive.
 
Here's the quote from the website.

Negligible electrical conductivity.
Arctic Silver 3 was formulated to conduct heat, not electricity. It is only electrically conductive in a thin layer under extreme compression.
(While much safer than electrically conductive silver and copper greases, Arctic Silver 3 should be kept away from electrical traces, pins, and leads. The compound is slightly capacitive and could potentially cause problems if it bridged two close-proximity electrical paths.)
Click to expand...
IIRC it said pretty much the same thing for AS2, which is what I've got here.
 
mboeller, c't was using a P4 2.26GHz, which may explain the lower scores on the CPU-bound flyby. I didn't think AA+AF would still be so CPU-bound, but it appears so....

In conclusion, all these numbers are confusing me, particularly since they don't include a 9500P for comparison. ;) I'll wait for "official" web reviews before deciding. As Tyan has just announced a 9500P board (see DigiTimes), I don't think we're in danger of running out of 9500P's anytime soon.
 
Ichneumon said:
One question I've not seen answered was weather the "Application" mode would use "bilinear AF" if bilinear was selected in the application... if it doesn't and it actually forces trilinear, then why do they call it Application in the first place?
Click to expand...
It does what the application requests, plus AF as set in the driver panel on top of that if the application sets no AF degree.
 
Xmas said:
Ichneumon said:
One question I've not seen answered was weather the "Application" mode would use "bilinear AF" if bilinear was selected in the application... if it doesn't and it actually forces trilinear, then why do they call it Application in the first place?
Click to expand...
It does what the application requests, plus AF as set in the driver panel on top of that if the application sets no AF degree.
Click to expand...

Ahh gotcha... so it simply uses the good quality AF at the settings in the control panel unless specified in the application, as well as bilinear or trilinear as specified by the app.

In other words... it really is Nvidia's "Quality" mode... But PR must play word games in control panel... *sigh*
 
Theres a question whether these 9600TX's are actually 9500PRO's clocked differently. If you look at the recent reports you'll see that Micron are said to have ordered 300,000 9500's - but the Micron adverts are all turning up with 9600TX's...
 
DaveBaumann said:
Theres a question whether these 9600TX's are actually 9500PRO's clocked differently. If you look at the recent reports you'll see that Micron are said to have ordered 300,000 9500's - but the Micron adverts are all turning up with 9600TX's...
Click to expand...
Medion, not Micron.
But the 9500Pro clocked 275/270 is faster than this 9600TX clocked 300/270. Either RAM timings are much worse, or this is a RV350 which is much more efficient than a R300 with 4 pipelines.
 
DaveBaumann said:
Theres a question whether these 9600TX's are actually 9500PRO's clocked differently. If you look at the recent reports you'll see that Micron are said to have ordered 300,000 9500's - but the Micron adverts are all turning up with 9600TX's...
Click to expand...

Yep, heard that too, but the benchmark-results are slower then the R9500Pro results despite the higher clockrate; and so I'm left wondering what's going on.
 
the R9600TX is not a RV350: http://www.pcwelt.de/news/hardware/30069/

my guess why it's slightly slower than a 9500pro even though it is clocked a bit higher would be slower ram timings (uses cheap tsop memory) and some disabled features, my favourite being Hier-Z.

I really have no clue why they call that 9600TX (math is complicated nowadays, not only do we have 2 == 4 < 3, but also 9000 < 8500 == 9100 and now even 9600 != 9600 == 9500).

edit: it is now officially confirmed: http://www.pcwelt.de/news/hardware/30084/
it says the name was an idea of ATI and Medion - they must have been smoking something hallucinogenic IMHO...
 
Ahh... that would explain the ridiculous multitexturing throughput we're seeing.

MuFu.
 
mczak said:
edit: it is now officially confirmed: http://www.pcwelt.de/news/hardware/30084/
it says the name was an idea of ATI and Medion - they must have been smoking something hallucinogenic IMHO...
Click to expand...
MuFu said:
Ahh... that would explain the ridiculous multitexturing throughput we're seeing.

MuFu.
Click to expand...

You mean smoking something hallucinogenic affects multitexturing performance?
Would this be the effect they called "flower power" some 30 years ago? :D
 
You must log in or register to reply here.

Similar threads

S
Memory bandwidth vs memory amount *spin off*
2 3
Replies
42
Views
36K
Kb-Smoker
K
S
Quantum computer to be presented the next week
2 3
Replies
50
Views
9K
davefb
D
S
Some speculation on Next gen consoles :) Very Verbose!!!
2 3 4
Replies
72
Views
11K
scificube
S
S
About NVIDIA, Futuremark and HardOCP
2
Replies
35
Views
7K
XForce
X
S
About NVIDIA, Futuremark and HardOCP
Replies
0
Views
2K
Scali
S
Back
Top