Little observation on Tom's hardware's review

[SteveC]

I'm not saying here that all of THG's review of the Parhelia is wrong, but I just noticed something which hints at proving THG does fudge results slightly

(I'm assuming here though that anand's XP2100+ is of very similar performance to THG's PIV 2.2Ghz in 3dmark2k1SEv330 for the fillrate tests - if this is not true, I'm talking out of my R-se )

Results from the THG jury:

Multitexturing Fillrate:

ATi 8500 1866.4
Nvidia GF4 2321.8
Parhelia 2478.0

Now, this shows the GF4 way ahead of the 8500 and Parhelia infront of both, but not by a massive amount.

Results from the Anandtech jury:

Multitexturing Fillrate:

ATi 8500 2048.4
Nvidia GF4 2324.4
Parhelia 2758.9

Now, this shows the GF4 ahead of the 8500, but by a smaller margin (ATi has gained 10%), the GF4 is pretty much identical and Parhelia infront of both, but this time, by nearly 20% instead of 5% or so as THG showed.

I dunno about you, but to me Anand has shown all three cards performing optimally, and THG has only shown the GF4 running optimally (hence the same result).

[Jake]

Or for some reason Parhelia perform better on Athlon XP's....

you b P BB? Pee Butty Boy?

Sorry Steve but your sig is just TOO silly.

Regs,

Jake

[SteveC]

Not just Parhelia performing better, the ATi too. The ATi seemed to also be shown in a bad light in THG's tests. (Comparing to Anandtech here, where the GF4 performs the same on both sites).

And leave my sig alone

[Guru]

Originally posted by SteveC
And leave my sig alone

OK Penis Butt Boy!

[SteveC]

Trying to be serious here - comparing all the more 'advanced' 3dmark results between what I get, Anandtech and THG get, THG consistently rates the ATi and P lower than they should be.

WHERE THG got their Pixel shader results from I have NO idea!


PS: With my sig, Parhelia Beta Boiz doesn't rhyme as well as P BB, does it

[Guru]

Tom used : NVIDIA - v29.42
ATI - v6.13.10.6094
Matrox - v1.00.25

Anand used : ATI - CATALYST 2.1
Matrox - v225
NVIDIA - v29.42

So Anand used newer M and A drivers right(?) and they both used the same N drivers!

PS: With my sig, Parhelia Beta Boiz doesn't rhyme as well as P BB, does it

Whatever you say PBB! We all know what you mean anyway!

[SteveC]

All online editors got the same drivers shipped with their Parhelia.

With respect tothe ATi case - I don't know what's newer/older and what's faster/slower.

[Jake]

Haven't you done you own tests to compare? Your XP1900 isn't all that far behind. Just OC the mofo. Run the tests yourself. Post some numbers.

You do have a P in da house, right?

[SteveC]

ralf - thanks!

Originally posted by Jake
Haven't you done you own tests to compare? Your XP1900 isn't all that far behind. Just OC the mofo. Run the tests yourself. Post some numbers.

You do have a P in da house, right?

I don't have to post numbers, from my own tests I get very similar to Anand's results, just a little less due to my CPU running a little slower.


edit: actually, that was assumed there, I just ran my own tests and get basically the same as Anand apart from the EMBM test for some reason I get over 20fps more than he did....

[Parhelia Pro256]

Gee, maybe a little <i>wine</i> and CHEEZE

[Kosh Naranek]

Originally posted by ralf
For one thing, THG:s 8500 runs at 250/250 according to the test setup.
Anands 8500 might be running at 275/275.

Regarding the P THG claim memory speed is 250, but I think that is a misprint, anyway Anand says 275.

I don´t know if Anand is testing with XP or 2k, but THG uses XP.

That said, if you look at the scores for pixel shader tests you will see very high discrepancies in the scores for 8500 and P between Anand and THG.

edit: I see you already noticed the latter.
Guru: the driver versions are the same (really)

THG uses W2K not XP.

[Indiana]

Yes, THG actually used the newest ATI drivers (the Catalyst package contains the 6094 version). But I have complained before that he DOES use an LE version of the ATI @250/250 and for comparision takes the highest clocked GF4. Besides his R8500 results are generally a bit low (I have a R8500 myself), so who knows what he did to get lower results out of the Parhelia as well. Didn't he use a slower clocked bulk version there as well.

[R.Carter]

Originally posted by ralf
Regarding the P THG claim memory speed is 250, but I think that is a misprint, anyway Anand says 275.

The bulk Parhelia has a mem clk of 250, while the retail has 275.

So it could be that THG is using the slower bulk Parhelia card 200/250, while Andand has a retail card 220/275.

[RedRed]

It doesnt really matter - he might not have used optimun settings on the motherboard or other peripherals - he wanted it to fail- its that simple!

Wait until we can buy one of these things.... we show some decent tests!

[RedRed]

btw - Haig, where is my Beta Boy application form!