I'm a system builder in Australia and therefore find myself at Toms Hardware quite often, checking out all the latest reviews and Benchmark tests.
I'm starting to get a little annouyed at the P4 vs Athlon benchmark tests here at Tom's. They simply do not hold any real world relevance. No-one I'm selling PC's to are buying P4's with RDRAM. My high-end systems are always Athlons with DDR333 or P4's with DDR333. I think you'll find this is the trend the world over, so why aren't we seeing benchmarks comparing the CPU's in this configuration?
If you want to include P4 with PC-1066 in the benchmarks then fine, I don't care, but please PLEASE also show everyone how the P4 performs with DDR333(PC-2700).
From my experience with my own bechmark testing the Athlon is still a pretty comfortable leader in the CPU stakes when both platforms use DDR333. I think the world should be better informed about this.
It would be more relevant to show the P4 using a DDR setup as that is far more common. I suggest you head over to Aces. He usually has a very broad range of system configurations and benchmarks. Tomshardware isn't the only site on the internet you know.
"We are Microsoft, resistance is futile." - Bill Gates, 2015.
>Comparing AMD/DDR to P4/DDR is like comparing AMD/SDR
>(PC133) to P4/DDR. Not a fair comparison at all..
No one is buying athlons with SDR; yet many, if not most P4 systems are being sold with DDR. So I think adding DDR P4 benchmarks is a very valid question. IM not saying they should ban RDRAM benches, as they show what performance you can get if money isnt an issue, but I agree with the original poster that adding DDR Ram benches is meaningfull.
= The views stated herein are my personal views, and not necessarily the views of my wife. =
You'll get your wish in a few days when Tom switches to using Granite Bay as the only DDR setup to really offer the maximum performance. With 4.2GB/sec Dual Channeled PC2100 which is cheap now to buy, I do not expect RDRAM to win at all if the SiS 648 or the i845PE with DDR333 are able to match or closely match the PC1066 setup.
Unfortunatly if you were directing this to Tom himself or any of his editors, you won't get anything from it, only David Stellmack and Omid come, but so rarely that it's for putting a discussion topic, and I tell you again, chances are 1 in 10K that this would be read by them.
*You can do anything you set your mind to man. -Eminem
While I agree it would be helpful to see cpu reviews use a variety of memory options, I have to say that the extra bandwidth PC1066 has over PC2700 does not enhance performance by much. In fact PC2700 has some advantage over PC1066 in latency and this can enhance performance in real world situations.
Here an article that compared various P4 memory options and shows that PC2700 can be competative with PC1066.
<A HREF="http://www17.tomshardware.com/mainboard/02q2/020520/ind..." target="_new">click</A>
In general PC1066 offers slightly better perfomance, but PC2700 is rarely far behind and can even edge ahead.
That article you have linked to just proves my point, if you know how to read the figures.
The only benchmark results in that article that show DDR333 close to, or very slightly ahead of PC-1066 are the ones that have ALL memory types very close together. In other words, the memory is not the limiting factor in those benchmarks. The bottleneck may well be cpu or something else.
In every bench test that shows a decent spread between the memory types (i.e. memory is the bottleneck) PC-1066 is clearly ahead on every occasion.
The fan I used for my k6-2 500 overclocked to 550 (don't laugh) was a big ass fortis copper heatsink, with some cheapo GlobalWin 60mm fan. It seemed to work ok. Of course, I did slap about 3 80mm on the little case I had it it. Without a nice fan/hs those things get HOT. That along with an old voodoo3 2000 made for a nice oven.
That Titan of yours is designed to handle an Athlon 1.4 which dissipates 65 Watts of heat nominally (more overclocked). The K6-2+ 450 only needs to dissipate 16 Watts. Your D5TB will be more than enough.
By the way, I used to use a HSF identical to the Titan TTC-CU5TB found on your link page. I used the Evercool <A HREF="http://www.nexfan.com/29/331.htm?404" target="_new">CUD-715</A> (all copper) but I switched to a cheap backup HSF, the <A HREF="http://www.nexfan.com/29/248.htm?404" target="_new">Galaxy GC21</A> (all aluminum) which actually cools better. (Sometimes simple really is better). Then I swapped in a quieter <A HREF="http://www.nexfan.com/43/425.htm?404" target="_new">YS Tech TMD fan</A> (don't run these off a motherboard fan header). This all cools (barely) my Tbird 1000 at 1500 Mhz. Talk about heat, though. My PC keeps my room warm.
<b>99% is great, unless you are talking about system stability</b><P ID="edit"><FONT SIZE=-1><EM>Edited by phsstpok on 12/18/02 03:14 PM.</EM></FONT></P>
Yes but it also showed that conventional DDR P4 systems are actually still behind AMDs, even 3.06GHZ ones, which surprises me a lot. That is what the majority of users here who got DDR setups will get. I thought it would be higher.
If you're willing to spend over $700 on a 3.06 P4, is money really an issue anyway?
The P4 works optimally with 1066 RDRAM, and should be benched that way. Not doing so is like not allowing an nForce2 in your Athlon benches.
While most of us would find a P4/DDR benchmark much more useful, it's a lot more fun drooling over an RDRAM setup we can't afford. Just like car magazines... out of reach, but it's still fun to know just how fast a ferrari can go.