Retro Anyone?

Comdrpopnfresh

Golden Member
Jul 25, 2006
1,202
2
81
We haven't used it in 5 years. It has an AMD Thunderbird 850mhz processor w/ a 200mhz bus. I didn't realize it could be overclocked until a bit ago...
I did a pencil mod on the L1 laser cuts to enable an open multiplier, then set the new base fsb clock to 133(needed to bridge the L1 points to do this as well). I then dropped the multiplier to get a final overclock of 1ghz, 266mhz bus, 133mhz Ram. I started @ 850,200,100. I can do a 1m calc in superpi in a smooth 2:20. It idles @ 36C, and loads @ 46 on a compusa brand copper heatsink with a 939 x2 stock fan on top. woot!

*EDIT* I changed the multipliers using the DIP jumpers on the mobo vs in bios- so if anything goes screwy I can reset bios and not have to re-change everything. I also changed the voltage from 1.7 to 1.775 (which droops to 1.83) using DIP jumps as well. The bus was changed from a base 100 to 133 using a jumper.
 

covert24

Golden Member
Feb 24, 2006
1,809
1
76
haha. i remember my AMD T-bird 1000mhz proc. thought it was faster than anything. It ruined my stepdad (at the time) iMac at the time.
 

Comdrpopnfresh

Golden Member
Jul 25, 2006
1,202
2
81
Funny thing is... I tuned my X2 3800+ down to 1ghz as well... It only completed each iteration of SuperPi 2 seconds faster (I attribute it to pc133 SD vs pc520 DDR, cache differences, and bus speed). I'd like to get my hands on one of the later Socket A boards, and put the same memory in each, and set all the speeds equal. Willing to bet they perform the same, seeing as K8 is a derivative of K7.... But then wouldn't that also mean that any barcelona product launched at the same speed as an X2 (speaking dual-core only here) would be roughly the same, because the K10/K8L is just a derivative of the K8, thus K7, and they have the same HTT, and in the case of the AM2 platform- memory?
 

f4phantom2500

Platinum Member
Dec 3, 2006
2,284
1
0
Originally posted by: Comdrpopnfresh
Funny thing is... I tuned by X2 3800+ down to 1ghz as well... It only completed each iteration of SuperPi 2 seconds faster (I attribute it to pc133 SD vs pc520 DDR, cache differences, and bus speed). I'd like to get my hands on one of the later Socket A boards, and put the same memory in each, and set all the speeds equal. Willing to bet they perform the same, seeing as K8 is a derivative of K7.... But then wouldn't that also mean that any barcelona product launched at the same speed as an X2 (speaking dual-core only here) would be roughly the same, because the K10/K8L is just a derivative of the K8, thus K7, and they have the same HTT, and in the case of the AM2 platform- memory?

nobody knows for sure about that k10 question, but i can tell you from first hand experience that a 2.2ghz athlon 64 stomps a 2.2ghz athlon xp, at least when it comes to gaming. it might not do superpi much faster per clock, i really don't know off the top of my head, but who cares? if you really care about calculating pi faster you'd get a more efficient program than superpi and do it in a small fraction of the time anyway.
 

Arkaign

Lifer
Oct 27, 2006
20,736
1,379
126
Heh, I have a spare box with a Tualatin P3 1.2Ghz @ 1.6Ghz with a TUSL2-C and 512MB PC150 Kingmax Memory. It is a very quick box, even with the old Ti4400 video card it still plays Enemy Territory really well. About as fast as a P4 2.2Ghz or AXP 2000+/2200+. Some of this must be due to the great L2 cache and high native FSB (not double/quad-pumped).
 

CTho9305

Elite Member
Jul 26, 2000
9,214
1
81
But then wouldn't that also mean that any barcelona product launched at the same speed as an X2 (speaking dual-core only here) would be roughly the same, because the K10/K8L is just a derivative of the K8, thus K7, and they have the same HTT, and in the case of the AM2 platform- memory?
K8, architecturally, is very similar to K7. HT and the integrated memory controller helped a lot with I/O, but for a purely compute-constrained task, they'll perform similarly. The Barcelona core has more significant changes that should improve the IPC (e.g. 128-bit floating point execution units, increased fetch bandwidth, etc).