Part Two - Possible Overclocking Of Socket A Processors With Contacts On CPU, Continued

We've looked into the coding of those contacts and could figure out the following so far:

The row L7 is clearly the area for the voltage setting. Compared to the list of the VID-settings of the Socket A processors it is obvious that the five bridges resemble VID4 to VID0 from left to right. A '1' in the table means the bridge is supposed to be OPEN, a '0' means it needs to be closed.

Row L6 represents the FID-settings, this time starting with FID0 going to FID3 from left to right.

So far so good. Unfortunately we've still not fully figured out the BP-FID settings, which you might remember from the first SlotA Athlon overclocking article . We know for sure that rows L3 and L4 represent the BP_FIDs, but since there's no list in the Socket A data sheets, we are not completely sure about the settings of them yet. We will of course come up with an article as soon as we've figured that last bit out. So far we know at least the settings for 650, 700, 800 and 1000 MHz, since we've got those CPUs. You can see these settings in the picture above.

Well, that's all fine, but what really kept me from publishing this finding so far is the problem to actually open one of the bridges. I guess that pretty much none of us has got a laser powerful enough to open up the bridges on the processors, and if you think that you can easily scratch them through then I need to tell you that you are sadly mistaken. I've tried several tools and so far my Socket A CPU's refrained from getting any bridge opened. So here we go, we are close to solving the riddle but not there yet. That's why this article is a bit premature, but after all the race is on now. With more people working on this issue after the publication of this article now world wide (even if it's just to beat Thomas Pabst) we should have a solution to it soon. Main thing is that every overclocker will finally benefit from the findings.

Intel Finally Admits That SDRAM Beats RDRAM

Again I am a bit late, since on Friday I was informed by a friend from a well-known memory company that is no big lover of Rambus Inc. that Intel has actually published performance figures of i815 and i820. For the ones of you who don't know about those two codes, 'i815' is Intel's latest chipset for Celeron and Pentium III, which is actually using PC133 SDRAM. 'i820' is also known as 'Camino(gate)', the most embarrassing chipset ever released by Intel, which is designed for use with the much hated RDRAM.

Intel did not do us the favor of publishing the numbers of i815 and i820 side by side, but if you go to www.intel.com/procs/perf/ , you will find out that a Pentium III 933 is scoring better on i815 in almost each benchmark that Intel used.

I guess that my point has finally been made. We proved that RDRAM gets beaten by PC133 SDRAM already more than four months ago, in the article 'Showdown at 133 MHZ FSB - Part Two '. While we were made responsible for an, unfortunately only temporary, drop of Rambus' stock, we had to put up with abuse from people who believed Rambus' marketing, which accused us of wrong benchmarking. Now Rambus will have to accuse Intel of the same 'sin'. At the same time I am taking my hat off to Intel. Who would have believed that Intel would finally be honest enough to admit that Rambus is inferior to PC133 SDRAM? Is this now an obvious sign that Intel backs off of Rambus? I sincerely hope so. If you ask me, Rambus doesn't deserve it any better. One Rambus employee, who I had the questionable honor to meet in Taipei last month, had actually the nerve to make Intel responsible for the bad performance of RDRAM in an email he sent to one of my editors. I consider that rather ridiculous! I was very tempted to publish this email, but with Rambus running to their lawyers when they are under pressure, and after consulting our own new in-house lawyer, I decided to be careful. It's a shame though! You would certainly have got a good laugh out of it. We old-fashioned Germans have an old saying that goes once you're under pressure, you shouldn't piss off the few friends you've got left . I doubt that Rambus has any friends whatsoever, but, besides Toshiba and Sony, Intel has certainly been the by far most important supporter of Rambus. Pissing off Intel by accusing them of bad engineering was probably not very wise. Now Rambus has got to pay the bill. Bye bye guys! I can't say it was a pleasure having you in the IT-business in the first place! Go with God, but go! And take your lawyers with you!

NVIDIA And Gainward - A Touchy Relationship

You certainly remember the review of Gainward's brand new 3D card with the GeForce 2 GTS chip. This new card was able to beat every 3D card out there by using 5 ns DDR memory, clocked up to 472 MHz. While many of you are waiting for the release of this graphics card NVIDIA seems rather unhappy. Here's the skinny:

That does obviously not sound very good and I wonder if Gainward has already destroyed the new relationship with NVIDIA before it really began. I am sure that I will receive a statement from Gainward about this issue soon too.

Celebration! We Broke The 20 Million Barrier!

Last but not least I think it is time to announce our new record. In June we actually served over 20.7 million html-pages! This is quite an achievement if you consider that this month had one day less than May, in which we had been able to make 19.8 million pages. Once I get all my 20 world wide employees gathered in one spot we'll make a big 20 million party and I am also planning to do a give-away for you readers, to express my gratitude for your continuous support of this website. We'll try our best to make this website good enough for you to score another record month in July 2000.