The Inquirer-Home

Hack to activate Hyperthreading on some non-HT Pentium 4s

Letters and killing time
Tue Mar 18 2003, 21:07
THE ARTICLE ON A SECRET hidden inside each Pentium 4 drew some fascinating letters. Not least someone who claims to have achieved the trick of switching on Hypertransport on a non-HT processor...

I have a gigabyte 8IHXPr2 mobo with 2.8 P4, and when I first flashed to new HT bios it booted up no problem and then when in windows I noticed win2kpro saw 2 procs, so I ran some benchmarks and indeed win2k pro saw HT as being enabled on my proc, well I let it run folding for a while and then I had a power outage and when the computer got rebooted the HT was gone, didn't show as duel system anymore, the second proc showing was gone. Another person did the same with 2.66 and it did same for his and he has been running it like that for months now and it works as a HT system, even though the 2.66 is not HT enabled. For the first boot and only first boot it seems to work, but as soon as ya restart it is gone. I am gonna attempt a flash back, and then flash up to see if it will do it again.

Email address supplied

alt='scissors'

It seems if you have a HT compatible motherboard without the HT bios that once you update to the HT bios and enable HT the processor will be recognized as an HT proc and function as such but once the computer is turned off or rebooted the HT is disabled and can't be turned back on as far as I know. There should be some way to work around this and keep it enabled. I've only done this on Gigabyte's GA-8IHXP motherboard but I've heard reports that it also works with granite bay mobos.

Email address supplied

alt='scissors'

Some quick calculations could mean spending far too much on computers to crack open the Xbox...

Unless I am mishtaken, this project will be 2^192 times harder than RC5's 64bit contest. That project took 1757 days. Without factoring moore's law, with the same ammount of CPU power as the RC5 project, Project-X should take 11,028,867,749,074,400,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000 days.

You know what that means? I'll need to install the client on 2^192 computers!!! AMD, do you take direct deposit? I would like to buy 6,277,101,735,386,680,000,000,000,000,000,000,000,000,000,000,000,000,000,000 Hammer processors.

Email address supplied

alt='scissors'

People are expecting too much from drivers, especially some of yesterday's letter writers...

I agree, I feel that people expect way too much out of drivers now. I update my drivers all the time though, because the bugs they fix and the updates they provide are worth far more than the *maybe* and *coincidentally* few slowdowns that *may* or *may not* occur in the process. OH WELL! Updates to make something work BETTER are worth a few slow downs regardless.

But yes, I agree….this driver crap has really been blown out of proportion. But hey, look at it this way. At least ATI releases driver updates that do in fact do something :)

alt='scissors'

In the letters to you regarding "Wild speculation on Opteron" a fellow named GEORGE states all is not well in VIA land and bombs them for CPU utilization on the EIDE controller. Stating that his Adaptec RAID controller got things under control at only 2-3% as opposed to VIA's horrible 35-40% usage. You might want to tell George to do his homework before he opens his mouth. The adaptec raid controller he is referring to has a HARDWARE controller on it (CPU of sorts) that relieves the CPU (Intel/AMD main cpu) of its duties.
The high usage of ALL onboard motherboard EIDE controllers (VIA/NVIDIA/INTEL whatever chipset) is because they are SOFTWARE controllers basically, and rely on the CPU to control all of its transactions. Traffic control (in lamens terms) is done by the MAIN CPU, not the controller on the chipset.
This is NOT the case with many EXPENSIVE adaptec controllers. They have a CPU of their own to do this job (an Intel I960 in this case). A QUOTE from their own website with product details on the 2400A:

"The Adaptec ATA RAID 2400A card is a microprocessor-based RAID solution for PC server environments employing ATA hard drives."

You don't get a dedicated microprocessor in your chipset (on ANY motherboard with EIDE) for this purpose. So send a msg to GEORGE letting him know his claims of your VIA BIAS at the INQUIRER (he all but said you were bought by VIA ads) is a bunch of bunk. Or he at least needs to dig up some better BIAS articles. He should also read some more motherboard reviews. Almost all of the motherboards (Quality names anyway...Asus/MSI/Abit/Epox etc) put out today with VIA/NVIDIA/INTEL/SIS chipsets are pretty damned stable. Even the crap brands are doing pretty good these days. Its rare when you read a review that says "X" board just sucks and crashed repeatedly (only BETA reviews do this here and there). While I'd agree the VIA/AGP fiasco on AMD based systems (and some Intel/Via solutions too) a long time ago was a major problem for a while, its long over and george needs to get over it.

I wonder if george even read the product literature on that card before he bought it...LOL Only a fool pays $300+ for a card and doesn't even know why he did it. That card is great, with 4 seperate MASTER EIDE channels, support for cache memory etc. But there is NO comparison to onboard EIDE controllers. They are two TOTALLY separate worlds. One you get on a motherboard that only costs $50-150 (with everything included, sound/lan/video/firewire etc) and the other is $320 just for ONE PRODUCT, that only controls your drives. What makes him think an "el cheapo" motherboard can keep up with a $320 hardware raid card? Thats like saying all is not well in NVIDIA land because my Onboard geforce 4mx can't keep up with my Radeon 9700 PRO. Who would expect it to keep up? I guess the same fool I mentioned above :) I hope he's at least running that card in a server or something quite taxing on a desktop PC. For the average home users this card is useless and you'd be better off with another 512meg, an extra 120GB drive and another $100 on your videocard.

I digress...
Aaron

µ

 

Share this:

blog comments powered by Disqus
Advertisement
Subscribe to INQ newsletters

Sign up for INQbot – a weekly roundup of the best from the INQ

Advertisement
INQ Poll

Heartbleed bug discovered in OpenSSL

Have you reacted to Heartbleed?