View Full Version : The new 1.53 palimino in sept question
Myself and I
08-10-01, 12:54 AM
Will this chip have a tottally different stepping than the new 1.33 and 1.444 which is ayjia or something.
Right now i am running a 900 slot a overclocked to 1000mhz on a old slot a motherboard. I think sept or middle of oct will be a good time to upgrade for me. Also what do you think the overclocking capabilities of this chip will have. Since it is the new palimino one might think it will be the next big overclocking chip
thanks
I've just read a review of a AMD Palmino 1.2 chip and the overclocking capability of the chip was poor. They could only overclock it to 1.33 stable and could only manage 1.4 with an new 400Watt power supply. I hope the new 1.53 chips are not the same
......
the little T-bird that could!
Pitspawn
08-10-01, 07:45 AM
Palminos are a very different type of chip to your average Thunderbird. Main differences are TLB (Table Lookup Buffer or something like that i dont know) which improves general performance by 10-15%, and PowerNow! PowerNow means a lower power consumption which means easier to cool. Another cool feature that i havent read anymore about in a while is cpu multiplier change through interrupt request (Like in windows!!). So basically software that could detect cpu temps and modify your clock speed if temps are too high or low enough to overclock etc. This process is called clock throttling. Because of TLB and cache all Palminos run faster than TBs meaning a 1.4TB Palmino runs more like a 1.55Ghz TB.
I cant wait for Palmino!! Unless my overclocking gets any better after I buy my new PSU, I will be buying one!
SickBoy
08-10-01, 07:58 AM
A TLB is a Translation Lookaside Buffer. It's a tool used in the system cache (which has not always been on-chip or on-die) to more accurately allow the processor to predict what instructions it will be executing next. The idea is, if the processor knows what it will need and what it will not need for the next X number of instructions, it can keep what it will need in the faster cache memory and send what it doesnt out to system RAM or to backing store (HDD). It does add a little burden/overhead to the whole process so how effective it is relates directly to its hit rate (how many times out of 100 or 1000 it will have the instruction it needs in the cache). An ineffective TLB is useless and is a simple burden of resources, while an effective one is a valuable tool for instruction execution.
So what I am saying is this: Say you magically "added" a TLB of greater size to your TBird. If it was an effective one, yes, you'd see a speed increase in certain benches and apps (probably Sandra's CPU test and the Multimedia test are ones that would be affected). If it was an ineffective one, you'd see no change or possibly an actual speed decrease in certain apps and benches.
Conclusion: Don't automatically assume because a processor has a TLB that it will automatically be faster, clock for clock, than an equivalent processor without one or with a smaller one. There's a lot in the implementation - it's the same concept as to why a 1.4 GHz P4 with RDRAM pales in comparison to a 1.4 GHz Athlon with DDR. It's all in the implementation.
SickBoy
Pitspawn
08-10-01, 08:01 AM
Thanks for correcting me [oc]SickBoy, I thought I was wrong :p
The palmino should overclock like a b*st'd on fire though from less power consumption. I look forward to the pleasure :)
Myself and I
08-10-01, 10:53 AM
Do you think a good 300 watt PSU like my aopen will be very effective? I know that they have 400+ out there, but i feel my 300 watt is really good right now , plus its actually a p4 psu comaptable
vBulletin® v3.8.7, Copyright ©2000-2012, vBulletin Solutions, Inc.