Ok this latest article had me reading until I got to the "Test Setup"
Who's the Genius that thought it was a great idea to talk about how SiS has a
license and is a main competitor to the newest VIA chipset release...
Yet, you Exclude the 645DX?!?!
You must be smoking so really good stuff...
well atleast you got 5 or so page hits, but include the info that is important to the
subject and you'd be amazed that people actually read the whole article when
it's balanced and well thought out...


someone get their thinking cap on, Please!
Craig

P3-700@933|G400DH|Asus P2B Rev.1.10 /w MSI slocket|128Mb PC100|SB PCI64|20Gb IBM 75GXP

talkin about Toms hardware?

-DAvid

-Live, Learn, then build your own computer!-

Yes.. the latest article posted on the front page


Craig

P3-700@933|G400DH|Asus P2B Rev.1.10 /w MSI slocket|128Mb PC100|SB PCI64|20Gb IBM 75GXP

A site that knows how to do a review..
they brought in all the new-comers Intel has, as well as the newest
Via solution... They didn't have a SiS546DX so they used a SiS645
which isn't optimal, but as the reviewer said it'll be pretty close..

and the SiS645 is faster...


Craig

P3-700@933|G400DH|Asus P2B Rev.1.10 /w MSI slocket|128Mb PC100|SB PCI64|20Gb IBM 75GXP

The SiS 645DX will be significantly faster than the VIA chipset if you run it with DDR400. The P4S533 supports DDR400 without overclocking. Therefore, it's safe to assume that the P4S533 would've been much closer to PC1066 in performance.

<b><i>"A penny saved is a penny earned!"</i></b>

Talking about someone at THG getting their thinking cap on, let's take another look at the Test Setup. Specifically the line:
<i>4x 128 MB RDRAM, PC1066, 400 MHz, 32ns, Samsung</i>

What are the <i>two</i> things that stand out?
1) PC1066 at <b>400MHz</b>? I <i>seriously hope</i> that this is just a typo and they <i>meant</i> to say PC1066 at 533MHz. Otherwise they're underclocking PC1066 RDRAM to PC800 speed!

2)<b>4x 128MB</b> means that they were running <i>four</i> RIMMs. Further, at 128MB, chances are high that these were double-sided RIMMs. (Notice how THG <i>still</i> doesn't specify the density of their RIMMs?) We all know that RDRAM's latency increases according to the number of chips actually involved. <i>SO</i> by running low-density RDRAM (twice as many chips) and <i>four</i> RIMMs of it at that, THG is <i>maximizing</i> the latency of the RDRAM system. Had they meant to actually show the maximum potential of a RDRAM system (instead of the maximum latency of one) they <i>should</i> have run it with two high-density 256MB RIMMs instead.

Clearly, this review was biased against the RDRAM platforms just by the hardware used alone.

Other than that though, it was nice to see them leave out the millions of stupid comments under the benchmarks. Just giving us the data was a nice change of pace. It's better to say nothing at all than to say something stupid.

<pre><font color=orange><b>du hast den Sweater verkehrt an</b></font color=orange>
Oh my!</pre><p>

Whatever happened to SiS 648?

I refer you to <A HREF="http://www.tomshardware.com/mainboard/02q2/020501/ddr400vsrambus-06.html" target="_new">http://www.tomshardware.com/mainboard/02q2/020501/ddr400vsrambus-06.html</A> and to <A HREF="http://www.tomshardware.com/mainboard/02q2/020514/p4x333-08.html" target="_new">http://www.tomshardware.com/mainboard/02q2/020514/p4x333-08.html</A>.

The SiS 645DX with DDR333 at CL2 scored 313.5 FPS in Q3 "Demo001". The i850E with RDRAM1066 scored 381.6 FPS. So with DDR333 at CL2, the SiS 645DX was 82.15% of the i850E's speed.

The VIA P4X333 with DDR333 at CL2 scored 327.3 FPS. The i850E with RDRAM1066 scored 355.2 FPS. So with DDR333 at CL2, the VIA P4X333 was 92.15% of the i850's speed.

I am comparing percentage of speed between the DDR333 platforms and the RDRAM1066 platforms under the same benchmark to aleviate any qualms about FPS gains or losses from CPU, video card, etc. differences, as in each case the RDRAM platform would have recieved the same performance hits/benefits as the DDR platforms would have from each individual review.

So we can see that the VIA P4X333 scored a 92.15%, where as the SiS645DX only scored an 82.15%. That is a 10% difference in favor of VIA. In fact, the SiS 645DX with DDR400 at CL2.5 scored 84.54% of the i850E with RDRAM1066's performance, which is still 7.61% <i>less</i> than the VIA P4X333 with DDR333 at CL2.

In other words, the performance of the SiS 645DX is sorely lacking compared to the performance of the VIA P4X333. I don't think that DDR400 at CL2 will be enough for SiS with DDR400 to overtake VIA with DDR333. The performance percentages just don't support it making enough of a difference. So unless SiS puts out some greatly improved drivers or a significant change is made to the motherboards using the SiS chipset, I can't in any way agree with you in your statement of "<font color=green>The SiS 645DX will be significantly faster than the VIA chipset if you run it with DDR400.</font color=green>". The benchmarks so far say otherwise.

We've seen cases of DDR266 platforms out-performing poor quality DDR333 platforms. I think we now have just seen a DDR400 platform whooped by a DDR333 platform.

Also, in the <A HREF="http://www.tomshardware.com/mainboard/02q2/020501/ddr400vsrambus-06.html" target="_new">http://www.tomshardware.com/mainboard/02q2/020501/ddr400vsrambus-06.html</A> article, we see mention of 'future chipset 1'. In that benchmark, THG already shows DDR400 performance. Yet neither of the mystery chipsets were running with DDR400. So it makes me think that these mystery chipsets therefore don't support DDR400. That leaves them to likely be either one of Intel's new 845 revisions, or VIA's P4X333 as 'future chipset 1'. This mystery chipset ran DDR333 at CL2 at 95.73% of the performance of the i850E with RDRAM1066, which is very close to the 92.15% that we know the VIA P4X333 just scored. So I suggest that either 'future chipset 1' <i>was</i> the P4X333, or else VIA's chipset has almost identical performance to Intel's latest and greatest 845.

So as much as I've vastly disliked VIA in the past (because of all of their damned bugs and crappy performance) I have to say that their performance at least has improved significantly. (Supposedly their bugs too.) So maybe VIA really is starting to shine. Certainly, as of today, I'd reccomend VIA over SiS. That much is for sure. I'm not sure if I find that funny or sad...

<pre><font color=orange><b>du hast den Sweater verkehrt an</b></font color=orange>
Oh my!</pre><p>

As far as I know, still marketting hype and nothing more. Since it is a SiS chipset though, one has to wonder if it will have their ever famous performance (or lack thereof).

Here's to hoping that they can learn and tweak before release.

<pre><font color=orange><b>du hast den Sweater verkehrt an</b></font color=orange>
Oh my!</pre><p>