tomshardware review online!

[flaimo]

not linked on the frontpage yet, and only the german version so far, but i guess you'll find your way through to the benchmarks..... typed in 020625 at the end just for fun and there it was....

not bad for a first post

[Ant]

Nice find

[GNEP]

How on earth did you find this flaimo? Insider info???

Like it... shame my GCSE german is not up to much.

gnep

[DrDS]

Not so good

[flaimo]

translation

[GNEP]

Looks like it's being held back by the pixel shaders - in 3DMk2k1SE330 (!) all the other tests show P performing (very) strongly. Interesting though that 16x FAA shows clearly better quality in screenshots than a GF4 Ti4600.

No "proper" benchmarking though - e.g. minimum framerates etc.

I tend to beleive the BBz w.r.t. "smoothness" of play - these game numbers all seem to be max/avg fps. whereas comments here have demonstrated other conclusions with regard to relative speed. Of course all this could be in the text for all I can read!

gnep

[xortam]

Is it still up? I was in the process of trying to archive it and its hanging.

[cjolley]

Seems to be down already.
chuck

[flaimo]

just overrun by visitors right now i guess...

[Znucke]

It's up, but it's really slow...

[Jon P. Inghram]

Originally posted by Znucke
It's up, but it's really slow...

What, the site or the Parhelia?

[xortam]

Got it archived now. Got to run it through Bable Fish.

[GNEP]

From the cache.

I will summarize then - the benchmarks.

Cards: GF4Ti4600 and Radeon 8500
Games: Giants, Max Payne, Q3A, Aquanox, Commanche 4, Dungeon Seige, Jedi Knight 2,
Resolution: 1024x768, 1280x1024 and 1600x1200, no AA or aniso I think

All results pretty much showed Matrox at the bottom, or just beating R8500. GF4 in the lead every time. I think they were average framerates (hmmm.....) not min or max.

German not good so they might have talked about smoothness of play but I wouldn't know any better.

3dMark 2001 SE 330: again, no AA. At the 3 resolutions above, 7201, 5633, 4366 respectively. As far below R8500 each time as it was below GF4.

There are also benchmarks against the GF4 only looking at AA and aniso. Only with both 16x FAA and max aniso settings does the parhelia match speeds. I will say this though - the screenshots look a damn sight nicer with the parhelia. But only when you download a 15 meg file with PNG formats in. The other shots are shrunken jpegs, so you can see jack sh*t. From my reading of the review, I would not be dis-heartened - but anyone else just looking at the pictures and graphs would straight away think "For that kind of money? No way. I will get a GF4," and then go and buy the 4200.

Oh well, I guess until we see some more in depth review (like min-max numbers from Serious Sam benchmarks) with more emphasis on picture quality up-front, this is what we should expect.

If the site is still down, I can answer more questions from the cache (got most of the pages - but only in the next 2 minutes 'cos I'm going home from work). I won't re-post any of the graphs as they will no doubt be copywrited.

gnep

[xortam]

Before we begin with the 3D-Benchmarks, nor some words to Parhelia. Matrox does not position the Parhelia as GeForce4-Killer. Instead of maximum bench mark numbers the Canadians refer to the very good performance with maximum image quality, more exactly said with the employment of anisotropischer filtering and edge smoothing (FSAA). Ago the true strengths from Parhelia would be to be looked for to. In order to become fair this beginning by Matrox, have we the bench mark tests into two passages divided. In first we regard the pure performance in the standard enterprise, thus with trilinear filtering and without FSAA, because many users still look primarily on the maximally possible Framerate. For $399 one should be allowed to expect there already something. In the further part we regard the image quality and the performance with maximum quality of Parhelia and compare them with a NVIDIA GeForce4 Ti 4600. We did not consult ATIs Radeon 8500 for these comparisons, since their anisotropische filtering by various driver-cheat is somewhat disputed and would have complicated the comparison only unnecessarily.

Value comparison The positioning of Parhelia at the market, thus not brachiale performance in the standard mode and but maximum representation quality, walk almost after a value comparison with the competition of NVIDIA. then it bench mark Parhelia show, are absolutely superior their GeForce4 Ti 4600 in the standard mode so mentioned. Standard means here: Only trilinear texture filtering and no edge smoothing (FSAA). By the image quality to increase one can let textures filter anisotropisch, which leads together with FSAA of too clearly increased image quality. Two questions are located in the area: Who supplies the better quality and with which performance. But in sequence. We turn first to the question of the image quality. Note: All Screenshots was made in a dissolution of 1024x768 with 32Bit colors. In order a correct impression of the quality to gotten, the Desktopaufloesung should amount to when regarding these pictures also 1024x768. We offer also all Screenshots in unkomprimierter form and in full dissolution for the Download, so that each can employ own comparisons. Also when regarding these pictures the dissolution in the own computer should amount to 1024x768

Result: Zwiespaeltige performance In things 3D-Leistung the Parhelia-512 leaves a zwiespaeltigen impression. A rather moderate performance in standard bench mark face partly very good results in the quality tests, even if the distance to NVIDIAs GeForce4 Ti is partly only very small. Positive also the high filling rate is in the Multitexturing enterprise. With higher GPU clock rate the Parhelia could be here still clearly faster owing to its 4 pixel pipelines than a GeForce4 Ti 4600. Something disappointing against it is cutting the Parhelia off in the pixel Shader tests of 3DMark 2001 SE. Here NVIDIA and ATI have clearly the nose in front. This point is particularly dramatic in view of the coming Gpu and/or. "VPU" generation of ATI and NVIDIA, which are already located in the starting holes and straight in things pixels Shader will add some at performance and feature (DX9). The new DX9-Features of the Parhelia-512, as for example DISPLACEMENT Mapping, becomes effective at present only into Matrox demos. The Parhelia-512 cannot keep up in things standard performance with NVIDIAs GeForce4 Ti. The argumentation of Matrox that the strengths of Parhelia lie in performance with maximum image quality, comes up not completely. With anisotropischer filtering the map cannot keep up with the GeForce4 Ti 4600. Points can it only with the qualitatively very good fragment AntiAliasing. With the Radeon the Parhelia partly supplies itself 8500 128MB von ATi in the standard tests a head on head duel, which in the view of the twice as expensive price ($199 opposite $399) however not straight appears honorable. A Top-3D-Performer is the Parhelia thus not, however the 3D-Leistung shown is quite very tidy. The 3D-Vergleich precipitates clearly in favor of of NVIDIA. We will compare the 2D-Qualitaeten of the Parhelia soon with NVIDIAs GF4 and ATIs Raedon 8500. On the paper Parhelia is there clearly superior. Together with the required 2D-Features, Surround Gaming, 10Bit Gigacolor and the good TV Out and signal quality the Parhelia can quite be a very good alternative to the established competition. One may be thus strained. It can be courage-measured already now that Parhelia without doubt will find very many friends. Who expected a super Athleten, the NVIDIA of the 3D-Performance-Thron falls, that however something will be disappointed. The 3D-Qualitaeten of the Parhelia lies in the detail.

[Kosh Naranek]

No Parhelia for me I guess ... I will save 200$ and buy a Radeon 8500 instead.

Come again M.