Pentium 4 1.8A Shoot Out

– Three different chips and motherboard chipsets tested– Jason

I was given the chance to assemble two P4 1.8A systems this week and I grabbed the opportunity to clear some of my doubts regarding different batches of 1.8A.

You may ask what are the doubts??

There are rumors which states that SL68Q is being produced with an optically shrink core, it operates at a cooler temperature and is more overclockable. These are my doubts.

As all of us know, P4 1.8A comes in two batches namely SL63X and SL68Q. Majority of SL63X were made in Malaysia while SL68Q originated from Malaysia and Costa Rica. Currently, I own a SL63X, therefore I specially hunt down two different nationality SL68Q to do a mini P4 1.8A shoot out.

Some information on tested P4 1.8A:

-----------------------------------------------
Production Code	: SL63X
FPO/Batch		: L208A955 (week 8)
Pack Date		: 06/04/2002
Nationality		: Malaysia
Purchase Venue	: Best Bargain


Production Code	: SL68Q
FPO/Batch		: 3219A012 (week ?)
Pack Date		: 05/30/2002
Nationality		: Costa Rica
Purchase Venue	: Costronics


Production Code	: SL68Q
FPO/Batch		: L222A573 (week 22)
Pack Date		: 07/09/2002
Nationality		: Malaysia
Purchase Venue	: Bell

---------------------------------------------------

These SL68Qs were bought on Monday 12/08/2002. During my search in Sim Lim Square, I found the following batches of CPUs. If you know those unknown batches, kindly share with us your findings. Perhaps someone could go check out those unknown shops (these are computer shops in Singapore):

1.	Chamoxa - unknown
2.	IO Data - unknown
3.	Hardware Place - unknown
4.	Visit - unknown
5.	Skylet - Can't be bothered to check them out.
6.	Fuwell - SL68Q, week 22, Malaysia
7.	Costronics - Refer to above information.
8.	Cybermind - Bundle price is too expensive!
9.	Bliss - SL63X, week forgot, Malaysia
10.	Best Bargain - SL63X, week 09, Malaysia
11.	Unlimited - SL63X, week 19, Malaysia
12.	Laser - No Stocks
13.	Voodoo Pro - SL68Q, week ?, Costa Rica
14.	BELL - Refer to above information.
15.	Superpet - Unknown (My guess is SL68X Costa Rica)

------------------------------------------------------------

Okie, that’s all to my findings. Hope it helps.
Let’s start the test!

Objectives: To find out whether SL68Q operates at a lower temp than SL63X and it’s overclockability.

Hardware Used For Testing:

All three CPUs ran 133FSB without any problem straight out of box. Therefore I started the test at 133FSB.
{mospagebreak}

Jason

The thermal probe is obviously not doing its job on ABIT BD7II. The temperature was offset by approximately 11 degrees! By touching the HSF on all three setups, my guess would be that both ABIT SA-7 and Gigabyte IGX were more accurate. The range of temperature, which all 3 CPU are operating at default voltage and stock HSF is approximately 36-51 degrees.

How accurate they were, I’m not sure. I’m still not technically skilled to accurately measure its temperature yet. But it doesn’t matters as results speak for itself.

As you can see from my findings, all 3 CPUs operate at almost the same temperatures. SL68Q might be optically shrunk in real fact (I don’t know) but its temperature is actually equivalent to SL63X. This is something that I am conclusive and positive about.

Surprisely, the Costa Rica chip cannot get past POST on all three setups @ 140FSB, default voltage. In terms of overclockablilty, Costa Rica is out of the competition. I was quite taken aback from this result as I thought all 3 CPUs should do well. Both SL63X & SL68Q(Malay) had no problems running at above settings.

The temperature is not included in this test as its not recorded down. Since the Costa Rica is eliminated, we have left with 2 chips to play with now.

{mospagebreak}

Jason

From above results, I found that:

1. Both CPUs fail to attain stability in all 3 setups.
2. Gigabyte IGX is not meant for extreme overclocking, hence both CPUs couldn’t boot.
3. With BD7II, both CPU fail at the several initial test of prime95.

From above results, I deduce:

• Both 1.8A requires 1.65V to attain stability @ 150FSB.
• Even increasing Vcore on SA7 & IGX does not produce any positive results. My guess for Abit SA7’s failure: Lack of feature to lock PCI/AGP. At 150MHz FSB, PCI bus is running at 37.5MHz, while AGP is at 75MHz.
• ABIT BD7-II is a good mobo to overclock your P4

Although Vcore was increased by 10% of the default voltage on ABIT BD7-II, Winbond doctor reports the voltage fluctuating around 1.58V.
{mospagebreak}

Jason

Since both SA7 & IGX held hands with the Costa Rica chip and were eliminated from the competition, there was only one motherboard left to play with.

SL68Q temperature lower than SL63X?

I don’t think so…

This is my conclusion for all above tests:

Throughout these tests, I always mislead myself that SL68Q (Malay) is gonna’ be a brilliant chip – but these tests proved me wrong! In terms of temperature, it operates at the same temperature with SL63X. This is something that I’m conclusive about.

In terms of overclocking, SL63X proved me wrong. It reached stability at 1.725V core @ 155FSB, but SL68Q (Malay) failed to achieve the same level.

One point I would like to share and note down is:

BD7II + SL63X @ 155FSB 1.725Vcore
CPU @ full load reports : 69 degrees
Motherboard temperature reports : 78 degrees
Condition : Stable

SL68Q(Malay) @ 155FSB 1.725Vcore
CPU @ full load report : 72 degrees
Motherboard temperature reports : 84 degrees
Condition : Prime95 failed

This is something I found strange. With two identical setups, both motherboards report different temperature. I did both tests at midnight and it’s got nothing to do with ambiant temperature. If you have observed closely enough, at test 14, SL68Q was operating 3 degrees more than SL63X! I tred re-seatting the SL68Q, but reports are still identical.

This was something that I cannot explain!

Both 1.8As did well to reach 2.7Ghz but I suspect it’s the heat issue that’s preventing both CPUs to advance past 160FSB. At 160 FSB, the HSF was real hot and I estimate it’s about 60++ degrees. I would love to lift both CPUs to higher grounds but my friend is still working on my custom made waterblock. I’m quite positive that via reducing more heat on the CPUs, both CPUs have high chances of hitting 3Ghz.

Wait! Don’t be mislead by my reports. I’m not saying that SL63X > SL68Q in terms of overclockability. In overclocking, it’s all a matter of luck. I do believe that I was unfortunate to hunt down a poor Costa Rica CPU.

CPUs of the same batches and production week yields different results. It does not indicate that via purchasing a renowned week/batch of CPU, it will guarantee you success in overclocking. It all boils down to luck.

All P4 1.8As should reach 133FSB at default voltage easily. Is SL68Q is a great CPU to overclock?

Results speak for itself.
Read and judge for yourself.

Jason