First Results: Celeron P4 Test

MrFrost · May 17, 2002, 02:07 PM

ASUS P4B rev1.02 - BIOS 1010 - 768M PC133 - Celeron 1.7G

I run Sisandra Burn-in App for 1 hour and 45 minutes; here is the benchmark:

(color reduced for size)
Not bad for an 88 dollar cripple-chip. And folks, this is at default 1.75Vcore. The thing is running pretty cool, too.

For Associative, here is a CPU-Z screen:

Interesting to note that NO L2 cache is detected! In fact, Intel's own cpu ID utility doesn't detect any L2 cache either!

Wonder what's up with that?

***Zardon*** · May 17, 2002, 02:09 PM

results are really impressive. have you gave it a good testing with everyday apps yet? windows response good etc?

MrFrost · May 17, 2002, 02:42 PM

Zardon -
Well, I've simultaneously run PaintShop Pro, SmartSaver compression, and SiSanda, while listening to an Audio CD. Then I played a video cd through PowerDVD while uploading, via Cuteftp, the SiSandra results to my server, with 2 browser windows minimized.
Right now, I have Dungeon Siege going in the background while I'm typing this.

Am running XP Pro Corporate, and everything seems strong and snappy.

The ease with which this chip has overclocked (to this point) rivals my best previous OC experiences with C300/C556/C600 and C1.0A.

On the bench, I tested the board at 110 the very first time I powered it up. Today, with a full rig, I immediately set it to 124...and you see the results.
Sometime this evening I'll go for my goal of 133 +.

Right now cpu temp is 101F, with a Arkuas 8586 hsf inside a Eagle(Skyhawk)4387 Aluminum case.
Neither of these is "Cream of the Crop" performers. The Asus P4B 1.02 is aging (but still an Excellent board).

All I can say is that, so far, I am not disappointed.

Mize · May 17, 2002, 02:57 PM

What were your Sandra memory scores?

Thanks,

Mize

***Zardon*** · May 17, 2002, 03:00 PM

having never really run Sandra on this processor your post made me very interested to see just how well an XP would perform in this test ...... the strange thing is that for some reason I couldnt get Sandra to work on my overclocked XP at anything more than 1.9ghz, but yet 3dmark and Pcmark would work fine ... ummmh, interesting.

I am very impressed with your results above and I certainly think the processor is a bargain for those on a tight budget and looking a bargain processor with excellent performance.

for anyone who is interested I took a screenshot of the sandra test after it had finished at 1.9ghz

MrFrost · May 17, 2002, 03:55 PM

Mize -
Mem bandwidth results were 947 MB/s with 94% efficiency. Must remember that it's only sdram, not ddr. (I've so much investment in sdram that I was too cheap to spring for ddr - that would mean a whole new mobo too.)

Zardon -
Man, those are some impressive numbers! I've not yet made it over to the 'darkside' because of the major (for me) expense of changing over to amd mobo, chip and ddr . Plus, changing chipsets (especially to VIA, so I've heard) can wreak havoc with the previous os installation.
Still....impressive numbers.

May 17, 2002, 04:03 PM

I enjoyed reading your posts about this processor and I think its a great price for a really good chip.

Things are looking rosy for the consumer.

keep in touch with any other benches you run etc.

Regards

***Zardon*** · May 17, 2002, 04:06 PM

I agree im impressed with those numbers for that celeron. its a bargain processor.

thanks for the posts, really good reading.

***Zardon*** · May 17, 2002, 04:12 PM

something worth reading http://www.driverheaven.net/thread.php?id=576

MrFrost · May 17, 2002, 09:02 PM

Okay, I ran 3DMark2001SE in a continuous loop for nearly an hour.
2862 3DMarks :o

Not anything great but then my vidcard is a VisionTek 32meg GeForce 256! (remember when those were Hot....and Expensive? Seems like a lifetime ago.)
Way outdated by today's standards but the price was right when I picked it out of the 'bargain bin' 6 months ago.

btw: Temps rose to 120F (48.8C) by the time the 6 loops were completed.

Also, tried 133fsb right afterwards but the system bailed on me. Looks like a latenight tweaking session.

***Zardon*** · May 18, 2002, 01:04 AM

it would be very interesting to see what kind of 3dmark you would be getting with a card like a Radeon8500 or a Ti4400/4600 in the rig.

Forge · May 18, 2002, 01:41 AM

Well, while I think it's great that your 90$ Celeron is doing 2.1GHz, I really think it's a poor choice, seeing as you can snag up a 2.6-2.7GHz 1.8A with *512K* of cache for just about twice as much. It *is* great that you can make a bottom end chip into a midrange one, but I wonder why you'd want to, when you can take a low/midrange chip and turn it up till it surpasses the top rated 'real' chip.

The headroom on Northwood is fantastic, and since they both require socket 478, I think we'd all be selling ourselves short to get worked up over a 128K Williamette.

Begin flames now.

www.tweakup.dk · May 18, 2002, 06:54 AM

Here in Denmark we also just is testing the Celeron 1.7Ghz cool running at 2.1Ghz without any problems.

The L2 cache is the same here, the computer don't register it, and the reason is that the chipset don't understand the smalla mound of cache.

If you are running CPUid then the number is still 0Kb, but down under i writes, cache speed "full" 2100MHz or 1700MHz... That means that CPUid actualy can se some cache, but not how much, and when running the tests, you can feel that they are there and i use, but just can been seen and identified yet. Thats something Intel and the other vendors is working on.

The Celeron P4 is pretty avsome to the prize, but the Northwood 1.6 @ 2.6 is Crazyyyyyy

www.tweakup.dk · May 18, 2002, 06:56 AM

Btw.. The new Intel Pentium 4 Celeron will be changed to Northwood technologi 0.13 micron and 256Kb Level 2 cache, and the the Old Pentium 4 Willamette is gone for ever with the Pentium 3, so only the Tualatin is back.......

JavaFox · May 27, 2002, 02:22 AM

I'm sorry, but... if you want to argue that the P4-based Celeron is impressive, that's fine -- but there was absolutely nothing impressive about the SiSoft SANDRA bench you posted. In the same way the frequency rating of this new processor has little to do with its speed, the SiSoft benchmark fails to translate to real world performance.

Forge · May 28, 2002, 03:24 PM

I agree 100% with Javafox.

If you want to go by Sandra, I'll put both the Athlon XP 1.9G and the Celeron 1.8G to sleep. Big margins.

If you want to go by real world, performance I'll put that Celeron down by even more, but that Athlon will give me a run for my money.

Comparing SiSoft benchmark results is about as worthwhile as comparing clockspeeds. You'll get the same results, too.

***Zardon*** · May 29, 2002, 12:08 AM

Quote:

Originally posted by Forge
I agree 100% with Javafox.

If you want to go by Sandra, I'll put both the Athlon XP 1.9G and the Celeron 1.8G to sleep. Big margins..

lol, lets see some undoctored screenshots of your test results, if you cant host them I will for you.

Forge · May 30, 2002, 02:22 AM

'undoctored' - Are you implying something there?

CPU/multimedia up once I get home in the AM. Any of the others you want? My crap ram makes my memory benchmark rather unimpressive (~2.2GB/s) and caps my 3Dmark2K1 score to about 13000, but otherwise it's a good rig. Any of the other benches reflect CPU power much, you think?

Forge · May 30, 2002, 02:28 AM

Just realized I already have Multimedia bench in a screenshot. It's got a bunch of other stuff on there, though (full screen), so it's a 200K shot. Please don't hit the link unless you really need to, and save it instead of reloading it.

(My ISP is gonna have fits if I get a /.)

http://forge.mine.nu/.images/p4.png

Feel free to mirror that on a faster connection, as my DSL isn't exactly 10mbit.