View Full Version : the bat computer 2.4C SL6Z3 part 2
If you remember in the last episode, I was able to finally hit 3.6 gig (300 FSB) with my new M0 stepping CPU, but I was not able to benchmark at that speed. I was able to run Sandra at 297 FSB though. Been burning in the system for a week or so now. Then Abit released their newest 1.9 version BIOS, so tonight I flashed the new BIOS and was playing around. To my surprise, I was able to benchmark at 300 FSB now (see screen shot). Don't know if it was the burn-in or the new BIOS or a combination of both.
http://mywebpage.netscape.com/batboybain/Sandra3600.jpg
Since I was screwing around and since the house was sort of cold with the heat turned down low (system idle temp was only 17 degrees), I decided to go for a max CPU-Z screen shot just for giggles. Holy guano, look what I was able to do! This is with my water cooled system listed in my signature.
http://mywebpage.netscape.com/batboybain/M0-cpuz-cpu-3700.jpg
Been researching the 3.3v power supply voltage mod. Think I might try it and see what happens. My 3.3v line averages 3.28v at 300 FSB when I was benching with Sandra. I'd like to see if I could run that a little higher. Supposedly this helps run the vdimm and VTT a little higher for a better RAM O/C too. I'm too chicken to try the mobo vdimm mod on my IC7, so this might be the next best thing.
shortbus_
12-09-03, 10:49 PM
nice job :eek:
i cant wait till mine comes.
For the record, that 3.7 gig attempt was way far from stable. The system crashed on me the first two times I tried to save the image... LOL. It's fun to look at the screen shot though.
For those that are wondering what I'm yapping about when I mentioned the PS 3.3v mod, here is a link that spOOL provided. I just found out about this very recently myself.
http://www.bleedinedge.com/guides/psu_3.3v_mod/psu_3.3v_mod_pg1.html
So you're thinking of modding your TP 550 then? I'm interested in trying this out on mine...
Yes, I'm serious about trying this mod. First, I need to take the PS casing off to double check to make sure there isn't already an adjustable pot inside. I'm thinking about trying this next weekend if I can round up all the resisters and pots.
Cool. I'd really like to try this out, but I'm really illiterate when it comes to circuits and, well... basically anything that has to do with modding electrical components. I just don't have the experiece I guess. I'm definately going to have to do a bit of research before I try this one out.
Detract
12-09-03, 11:41 PM
Congrats on hitting 3.6 batboy. Im gonna try the new bios tonight. Also im gonna try mod the 3.3v line. At the moment it's at 3.2v :eek: .
GigaHertzAddict
12-10-03, 01:21 AM
That's very cool.. I up to 3.7GHz myself, but it needs too much voltage to get there.. 1.85v in bios and about 1.725V at load.. Too bad the Asus overvolts at idle, yet it was more stable than a ABIT IC7
Yup, my 2.6c can do 3.7 with 1.8v too but i'm really scared.
Very impressive.
One comment though.
In that benchmark you INT bench is high but you FPU/ISSE2 are past their prime.
On my 2.4C I get the best FPU/ISSE2 is achived at 280 4258/7446, after that it goes downhill.
Why do the CPU benchmarks diverge at high speed?
This might be common knowledge but I only just realized, since I've been too busy playing with memory.
But sure it is cool having a high FSB and I'd love to break the 300 limit myself, though it is not looking likely at the moment.
Good luck!
reverse
12-10-03, 08:19 AM
Nice! :D
nice! :D. batboy, is that on 3:2 divider or 5:4?
It's with the 3:2 ratio which might explain why the benchmarks isn't higher. My RAM gives up using the 5:4 ratio at 283 FSB because I'm limited to 2.8v vdimm. I'd love to run this computer all the time at 3.5 gig if I could use the 5:4 ratio. That's why I'm going to try to do this power supply 3.3v mod.
deathman20
12-10-03, 09:48 AM
Oooh nice guess I'll have to try the new bios too and see what it holds for me.
Nice OC and good luck for the need for speed.
Jimbob7
12-10-03, 12:55 PM
Yeh get that vdimm nice and high :D
i must wait till xmas untill i can put my mobo chip and ram together:mad: :cry:
Jojo1971
12-10-03, 01:25 PM
**modding the PSU without modding the vdimm thru the mobo will NOT work...you got to do both....the mobo has it's own regulator so even if youre able to raise the PSU 3.3V line to 3.6, your vdimm will still be the same.. Now if youre just going to do the mod in the mobo , the highest vdimm you can get will be lower than what the PSU's 3.3 line output is..
get a low-output pointed soldering iron and a small tweezer and start doing it batboy.. we want to see your results...:)
Surfeit00
12-10-03, 01:57 PM
bat - did you use clockgen for your suicide runs?
OMG, i'm such a noob.....My 3.3V line has been running at 3.1V (tops) forever, and I didn't really pay much attention to it...
Well, my experience with a soldering iron is next to nothing, so its off to buy a new PSU?
Or what? How "extreme" does your OC have to be, for the (lacking) 3.3V line to br an issue?
climbski
12-10-03, 02:24 PM
Originally posted by Jojo1971
**modding the PSU without modding the vdimm thru the mobo will NOT work...you got to do both....the mobo has it's own regulator so even if youre able to raise the PSU 3.3V line to 3.6, your vdimm will still be the same.. Now if youre just going to do the mod in the mobo , the highest vdimm you can get will be lower than what the PSU's 3.3 line output is..
get a low-output pointed soldering iron and a small tweezer and start doing it batboy.. we want to see your results...:)
Absolutely correct .. you have to bypass the vdimm voltage regulation first with a resistor mod to the board...sorry if you misunderstood me earlier.
Bat,
NIce oc but I just hit 12x302@3624 with 3:2 chip is almost 1 week old..........posted at 315..............
jdmcnudgent
12-10-03, 11:43 PM
congrats batboy. nice chip you got there.:)
micamica1217
12-11-03, 12:51 AM
sorry for the realy late reply......nice OC batboy.:clap:
for the first time in a long while, I'm flashing my mobo to a new bios.
I think I'm still using the 1.7b bios.
anyway, I'm folding now, and my 3.3v line states 3.25-3.26.....
what do you think of this?
mica
congrats on finally getting there, can you prime at 300FSB?
Sorry for not replying sooner, been out of town for work.
Not Prime stable at 300 FSB, yet.
This lack of vdimm is starting to annoy me. Wish Abit would fix the Max 3 bug. I might just order an AI7 mobo and Mushkin PC3500 back level 2 RAM.
Surfeit, you are correct, my "suicide run" to 3.7 gig was with clockgen. The 300 FSB benchmarking was done the old fashion way.
I have an AI7 sitting on the floor headed back to where I got it. The vdimm/vtt both work flawlessly stable up to 3.2/1.6. However, the 865 chipset is noticeably slower than the 875. I lost 500 points on my cpu score in PCMark2002 while gaining 50 on my memory.
This Asus board seems a tad quicker than my max3(on a shelf, too late to return). Also, I've noticed with the vdimm volt mod, underload of memtest86 test #5, the vdimm drops from 3.1 and 3.2 to 3.05. If I set it to 3.05 it stays at 3.05 under load. This I'm relating to the 3.3 line of the psu or the mobo circuitry.
There is a volt mod out for the max3 which I may try since I'm probably never going to use the max3 again. I might try the 3.3 line mod this weekend myself. I picked up some trimmers at radio shack earlier tonight. 220 is the highest 1:1 @ 2-2-2-6 and 225 @ 2-2-3-6 with my current setup. I'm thinking 3.2-3.3 volts would yield 2-2-2-5 at 225-230, although the cpu might become my limiting factor.
flapperhead
12-12-03, 07:48 AM
Originally posted by batboy
Sorry for not replying sooner, been out of town for work.
Not Prime stable at 300 FSB, yet.
This lack of vdimm is starting to annoy me. Wish Abit would fix the Max 3 bug. I might just order an AI7 mobo and Mushkin PC3500 back level 2 RAM.
Surfeit, you are correct, my "suicide run" to 3.7 gig was with clockgen. The 300 FSB benchmarking was done the old fashion way.
congrats on the 300 fsb guy. i been working so much overtime i havent been able to play with my new m0 2.4c. have u considered doing a vdimm mod to the memory voltage feedback?Its a real pain in the a** but it really helps with increasing memory speed and stability
almost forgot to mention, as u probably already know , upping the agp voltage to 1.65 considerably helps the nb stability
I've considered the vdimm mod, but I suck at soldering, so I'm sort of chicken to try. I've tried upping the AGP voltage, but it does nothing for me.
Originally posted by batboy
I've considered the vdimm mod, but I suck at soldering, so I'm sort of chicken to try. I've tried upping the AGP voltage, but it does nothing for me.
Nice overclock so far, altough I can't feel very enthousiastic about the Max3. It's little to no difference compared to other solid 875 boards as far as I can see, proving the futility of the mosfet air duct etc. Ah well.
About that vdimm mod- not all mods require a soldering iron. It's sometimes also possible to use 'grabbers' for the mainboard contact points and connect the resistor and wires to these. Maybe that could be an option in your case too.
Detract
12-12-03, 05:41 PM
Originally posted by FIZZ3
Nice overclock so far, altough I can't feel very enthousiastic about the Max3. It's little to no difference compared to other solid 875 boards as far as I can see, proving the futility of the mosfet air duct etc. Ah well.
About that vdimm mod- not all mods require a soldering iron. It's sometimes also possible to use 'grabbers' for the mainboard contact points and connect the resistor and wires to these. Maybe that could be an option in your case too.
Unfortunately you have to desolder a small resistor in order for the vmod to work on the IC7. I did it myself 2 weeks ago, it aint easy because the resistor is so damn small. Lucky enough for me it worked and I can run the ram upto 3.15v, which means 220Mhz 2,2,2,5 100% stable.
gruvin2
12-12-03, 09:12 PM
Grats on a killer chip Batboy! keep burning in and let us know your max stable. As for vdimm mod i highly recommend it. It allows for much better timings @ higher FSBs then at 2.8volts. I took my ram up to 3.22volts and its still only warm to touch at load. Sounds like yours is a bit more involved then it was with my MB. Gluck and keep us posted :)
hepp- to answer your question about higher clocked CPU's giving diminished returns is simply the CPU or RAM having errors and having to spend more cycles to fix em. Hence the less total output. At some point the CPU is going to be less effective. Just takes time to find you Rig's "sweet spot".
Originally posted by Detract
Unfortunately you have to desolder a small resistor in order for the vmod to work on the IC7. I did it myself 2 weeks ago, it aint easy because the resistor is so damn small. Lucky enough for me it worked and I can run the ram upto 3.15v, which means 220Mhz 2,2,2,5 100% stable.
So you actually remove part of the motherboard... and it works better?
Originally posted by Detract
Unfortunately you have to desolder a small resistor in order for the vmod to work on the IC7. I did it myself 2 weeks ago, it aint easy because the resistor is so damn small. Lucky enough for me it worked and I can run the ram upto 3.15v, which means 220Mhz 2,2,2,5 100% stable.
Oh I see, well that's too bad then. :(
Detract
12-13-03, 04:31 PM
Originally posted by tio
So you actually remove part of the motherboard... and it works better?
The resistor is 0 Ohms and is merely a bridge. Adding a little resistance via a VR pot, increases the voltage to the RAM. It works because you can o/c the RAM higher than at stock voltages.
flapperhead
12-15-03, 07:53 PM
hey batboy, the question occured to me with your watercooling setup. have u tested the wcooler in comarison to an air cooled setup?or if not. Is it possible to estimate how much of an improvement it is over a hsf?
mamisano
12-15-03, 08:03 PM
Originally posted by Detract
The resistor is 0 Ohms and is merely a bridge. Adding a little resistance via a VR pot, increases the voltage to the RAM. It works because you can o/c the RAM higher than at stock voltages.
Yeah, that's what I have to do. I just stopped at Radio Shack today and picked up the essentials. Waiting for my 2.4C to come in from ZZF, then I will swap the processor and perform the mod. I hear that BH-5 chip Memory like a lot of voltage.
flapperhead
12-15-03, 08:16 PM
Originally posted by mamisano
Yeah, that's what I have to do. I just stopped at Radio Shack today and picked up the essentials. Waiting for my 2.4C to come in from ZZF, then I will swap the processor and perform the mod. I hear that BH-5 chip Memory like a lot of voltage.
make sure u replace the the 0 ohm with a 10 ohm resistor and place the input of the pot on the side of the new resistor closest to the windbond ic. just replacing the orig. resistor with a vr can cause voltage stability probs and or vdimm limited to around 3.15
Originally posted by flapperhead
hey batboy, the question occured to me with your watercooling setup. have u tested the wcooler in comarison to an air cooled setup?or if not. Is it possible to estimate how much of an improvement it is over a hsf?
Yeah, I was running this CPU air cooled at first and got 3.55 gig max Sandra stable vs my watercooled 3.6 gig max Sandra stable. So, not a huge difference. The air cooled heatsink is the same that's on my 2.8C in siggy.
Mr Gadget
12-15-03, 11:23 PM
Originally posted by flapperhead
hey batboy, the question occured to me with your watercooling setup. have u tested the wcooler in comarison to an air cooled setup?or if not. Is it possible to estimate how much of an improvement it is over a hsf?
Joe Citarella has compiled a lot of data on both air cooled and water cooled heat sinks. From his data:
Air Cooled (http://www.overclockers.com/articles373/p4sum.asp)
Water Cooled (http://www.overclockers.com/articles373/wbsum.asp)
It looks like the newer air cooled heat sinks have closed the gap in terms of thermal performance. The best of both systems seem to be in the range of .14 to .16 C/W. The water cooled system maybe quieter though!
The MCW5002 water block is .15 C/W as compared to the MCX4000 at .20 C/W that works out to 5C on a 100W System, which is probably not enough improvement to justify the cost difference.
jazztrumpet216
12-16-03, 12:01 AM
*falls off of desk chair...*
Sweet Jesus... that is one heck of an overclock. Congrats man...
Originally posted by Mr Gadget
Joe Citarella has compiled a lot of data on both air cooled and water cooled heat sinks. From his data:
Air Cooled (http://www.overclockers.com/articles373/p4sum.asp)
Water Cooled (http://www.overclockers.com/articles373/wbsum.asp)
It looks like the newer air cooled heat sinks have closed the gap in terms of thermal performance. The best of both systems seem to be in the range of .14 to .16 C/W. The water cooled system maybe quieter though!
The MCW5002 water block is .15 C/W as compared to the MCX4000 at .20 C/W that works out to 5C on a 100W System, which is probably not enough improvement to justify the cost difference.
That is not correct. The C/W ratings obtained under different testing conditions and with different equipment cannot be compared between tests!
Watercooling can still result in a better cooled system. The thing to keep in mind is that it will never go below ambient, because that is still the medium it uses to cool things.
Open evaporative watercooling can go below ambient, but the best bet for super low temps is phase-change. The big OC gains are also to be found almost exclusively with the latter type of cooling.
Mr Gadget
12-16-03, 09:44 AM
Originally posted by FIZZ3
That is not correct. The C/W ratings obtained under different testing conditions and with different equipment cannot be compared between tests!
I may have misread Joe's articles and tests, but I believe he used his Die Simulator (http://www.overclockers.com/tips263/index02.asp) to test the recent water blocks and air cooled heat sinks. According to his description of the die simulator he runs all of his tests up to 130 watts so I assumed that the results would be comparable since the thermal coefficient is the ratio of the 'change' in temperature to a 'change' in power.
I agree that phase change cooling is the way to go for all out performance, in my case I will probably stay on air until I have the money to invest in a Promethia.
To clear up any misconceptions, my watercooling rig runs at least 10 degrees cooler load CPU temp than my good air cooling (Swifty heatsimk with high speed Delta fan). Cooler temps mean more longevity. I was just saying it didn't help my max O/C as much as some people might think. The main reason I love my watercooled rig is the quietness. Ahhh, no more ringing ears caused from screaming fans. Nice!
deathman20
12-16-03, 12:28 PM
Well thats nice Batboy. I can say the same thing when I had my watercooling rig that it wouldn't run much higher then it was with a stock cooler at the time (got an extra 60mhz out of the CPU back then).
Watercooling is for quietness, coolness, and well just saying that "Yes I can watercool my PC :p"
Originally posted by Mr Gadget
I may have misread Joe's articles and tests, but I believe he used his Die Simulator (http://www.overclockers.com/tips263/index02.asp) to test the recent water blocks and air cooled heat sinks. According to his description of the die simulator he runs all of his tests up to 130 watts so I assumed that the results would be comparable since the thermal coefficient is the ratio of the 'change' in temperature to a 'change' in power.
Well, the rankings are not completely uniform. For some heatsinks, only "motherboard" measures are reported. For example on the best ranking SP94:
http://www.overclockers.com/articles798/
The .14 C/W is taken directy from the most aggressive mb test reported in the article. This is not a die simulator result.
For others, various die sizes are reported (probably the most extensive sort of review), like here:
http://www.overclockers.com/articles726/
Note the large difference between die's under otherwise equal conditions.
Finally, watercooling results may report die simulator results, but not (always) die size:
http://www.overclockers.com/articles886/
In sum, these are some of the reasons why you cannot compare directly from the rankings chart, and sometimes cannot at all.
vBulletin® v3.8.7, Copyright ©2000-2012, vBulletin Solutions, Inc.