If you have been fond of overclocking for a long time, then most likely you have heard about such a concept as “unsuccessful processor instance”. So, this is explained quite simply: an unsuccessful processor copy is a chip that is knocked out of the general statistics (for the worse, of course), not just by low overclocking capabilities, but simply by terrible overclocking potential, high base voltage, and a very hot temper.
Today I will continue my excursion to the processors based on the NetBurst architecture. The test subject that came to me for the test has the same Prescott core as the Celeron D 346, but its overclocking potential somewhat discouraged me. In all my conscious overclocking career, I have not yet seen such a flawed example based on the NetBurst architecture.
Test setup:
- Motherboard – Biostar P35D2A7
- Processor – Celeron D 336 (G1)
- Cooling – Cooler Master Hyper TX3
- RAM – at first 1GB DDR2 PQI PC6400 (5-5-5-24) but due to the “shamanistic dance” it was replaced with a 512MB Elixir PC6400 (5-5-5-16) bar
- Video card- Radeon X600 (450/450MHz)
- Power supply — Cooler Master 460W
- HDD – Samsung HD161HJ 160GB
Processor
The marking of our test unit is SL98W. Before us is a typical representative of the stripped-down Prescott core: the chip is manufactured according to the 90nm process technology, has at its disposal a modest 256KB L2 cache, and operates at a frequency of 2800MHz. Stepping is G1, and the country of origin is the Philippines.
The CPU-z utility gives us a little more information than parsing the marking of the chip: The base clock frequency of the processor turned out to be 2793MHz, it is obtained by multiplying the 133MHz bus by a factor of 21. Celeron D 326 has 256KB L2 cache, supports x64 instructions, and its base voltage the core power supply is 1.344V.
Overclocking
Well, if you still remember, I managed to squeeze as much as 4740 MHz frequencies out of the Celeron D 346 processor (for a screenshot) at a supply voltage of 1.76 volts. Full stability came at a voltage of 1.56 volts at 4144 MHz, which in itself is a fairly average result.
Nevertheless, the instance of Celeron D 336 I came across just shocked me: at a voltage of 1.71 volts, I could only achieve a pitiful 3779 MHz:
Any manipulations with increasing the RAM timings, overestimating the supply voltage even more (up to 1.850 volts (don’t do this)), changing the RAM, turning the cooler to different positions: Everything was in vain.
I immediately thought about the local overheating of the core, but looking into the section with sensors, and seeing 55 degrees, I was completely discouraged. After all, the same Celeron D 346, was able to boot, and at the same time respond to commands at a temperature of about 80-90 degrees, at a frequency of 4740 MHz!
The 90nm Prescott processor didn’t conquer 4GHz ?!
Except for the words “what the hell is this?” and deep bewilderment, I could not squeeze anything out of myself.
Conclusion
As it turned out, not all Intel Celeron D 336 processors with G1 stepping are equally useful – this is a fact. Yes, I came across a very, very bad sample. But don’t forget that not all Celeron D 336 chips are bad. My instance is most likely just a fly in the ointment.
