Coppermine PIII 500E – Next Great Intel O/Cer?

SUMMARY: Amazing off-the-shelf performance – I maxed out my SOYO 6BA+IV at 155 MHz giving me 776 MHz at 1.6 volts. The PIII 500E should rank with the C300a, PII 300 SL2W8, PII 333 SL2WY Week 37 and PIII 450 SL35D for overclocking ease.

The CPU in this review IS NOT an engineering sample – I bought this chip from High Voltage Computers and it came in a sealed retail box. This is a Malaysian PIII 500E SL3R2, FPO# Q944A007, MM# 825584, Version # A01304-001, Pack Date 11/6/99. It comes with an Intel heatsink which I did not test yet; I used an Alpha Socket 7 on a Soltek slotket SL-02A+. This is their Coppermine slotket which takes PIII 500 and 550s – apparently there is another version in the wings which will take higher speed Coppermines.

As you can see from the pictures, the chip itself is miniscule – it measure 3/8″ x 7/16″ and sticks up about 1 mm over the chip’s base. The back of the chip has a number of exposed components as well, probably capacitors. In practical terms, the size of the chip presents a real challenge when you mount a heatsink – it is very easy to mount a large sink (like the Alpha) in a way that it does not contact the CPU’s surface squarely. I took special care to make sure the Alpha was aligned evenly on the chip the second time I mounted it and noticed better cooling performance.

I think these chips may require better clips on heatsinks to ensure absolutely level mounting – I could not use my AquaStealth on this due to the lack of a secure mounting method. It was apparent from the CPU’s thermal diode temp that the AquaStealth must have been slightly off. I will have to make some kind of adjustable clip for this.

Mounting the 500E on the slotket with the Alpha (see pic below) required a slight modification – the handle of the ZIF socket is a metal one that made contact with the base of the Alpha. I carefully straightened it slightly such that the handle now rests below the Alpha’s base – very easy to do with a pliers but a necessary mod.

The test system was as follows:

SOYO 6BA+IV
Matrox 16 MB G400
Samsung G8 64 MB
Ensoniq PCI Sound Card
USR 56K
IBM 10.1 G HD

The SOYO picked up the 500E at boot, adjusting the voltage to 1.6. I was not sure if it would so I adjusted the voltage on the Soltek to 1.6, but this is not necessary – leaving it on AUTO is fine. I tested the various FSBs using SOFT FSB – I booted at spec speed and voltage and then varied FSBs. At each step tested I used Prime 95 to stress the CPU to a stable temp. The reported temps are using Motherboard Monitor readouts and reported CPU speeds. I also used SiSandra 99 to benchmark MIPS and MFLOPS at each stage – all of this data is presented in the following table:

Ambient temp 20C, system temp 23C, CPU temp at rest 19C

I pooped out at 155 MHz but I did not get a BSOD – Prime 95 failed but it was still running OK. Now looking at this table a number of things pop out:

o The temp gradient as I worked my way up FSBs is almost unbelievable – I did this twice to check out temps and got the same results. Either the thermal diode is wacky, or the flip-chip design is incredibly efficient at transferring heat from the CPU to the heatsink (I think the latter).
o I required no change in voltage to hit these speeds – NADA!
o The SOYO is one incredible board – I can’t say enough about its stability and ease of use.
o The Matrox G400 is one incredible AGP video card – at 155, this card is running at 104 MHz – I am frankly astounded that it did so.
o The Samsung G8 is also great – it ran over spec and did it without a hiccup.

I don’t know what alignment of planets made these speeds possible with this setup, but I can only tell you what I found and with what. This is a sample of one – as others buy the 500E and try it out, we’ll see if this is just a freak or not; I suspect it is not because the core is spec’d to hit 733 MHz, and I believe with the right board (whenever the 810s get bug free) and cooling I don’t doubt that 800 is achievable, and with super-cooling I’ll bet someone will break 1,000.

CONCLUSION

As I was testing the CPU, I was flabbergasted. I called my pal at PC Nut to see if I was hallucinating, and Humphrey verified that the Matrox G400 16 MB can in fact sustain very high AGP speeds and the SOYO is incredibly stable at all FSBs. I believe the 500E will not be around too long and if you are thinking of upgrading, this is way better than the PIII 450. I think Intel really did their homework on designing this CPU for heat dissipation and speed.

First impressions – an outstanding accomplishment.