Re:SiS and ALi don't produce quality anything.
on
AMD And THG update
·
· Score: 0
Your information is old. SiS used to be low quality, but their recent SiS735 socketA chipset is one of the fastest available. When using motherboards based on the sis735 with onboard sound/NIC, it is easy to build a FAST and CHEAP and STABLE system. I can confirm this because my room-mate has one. It doesn't crash (winXP). It is fast. It was cheap.
I just helped a kid put together a $450 system based on the 735, and it performs very well too.
Re:and in the end it doesn't really matter
on
AMD And THG update
·
· Score: 0
I don't know it would take that much work. I imagine throttling would be hard, but is it such a large step to go from being able to detect overheat to powering off itself like the P3 does?
If you're going to ask someone to risk their athlon, please demonstrate your pentium4. Specifically, I would be fascinated to watch a heatsinkless pentium4 boot into windows. That would be awesome. So... show me;)
Agreed. In Tom's video, the instant the heatsink came off, the thermal paste began to smoke and the system died. here, there was no visible smoke(or paste), and the system continued for a significant chunk of a second (maybe about 1/10th of a second or so). Something is severely wrong with one of the tests.
Even though it is possible, I would think Tom should have retested different CPUs and different mobos. You CANNOT generalize based on a sample size of one.
not exactly. in this case, intel would have to have the CPU realize, "oh, this is a standard benchmark i'm running. let me work at half the floating point accuracy so i can outperform the athlon"
Re:Subscriptions should add value
on
Slashdot Updates
·
· Score: 0
the one thing I dont like about that is
. Only subscribers can submit stories.
I think that the priviledge to submit should be given to everyone.
I didn't really like mandrake when i played around with it... does the default install not install binutils? i was trying to get the latest kernel installed, and it was a real pain. maybe i just dont do well with gui everything;)
What is the max delta T across these devices? If you want to create a fanless system using one of these, the heatsink on the hot side has to disipate LOTS of heat. that means it will get a lot hotter than room temp, so if these devices can only create a 30C difference, you might only get slightly below room temp on the cold side
I recall reading something on the web (was it here?) where they timed some linux compiles, and the dual tbird was more than twice the speed of the single.
a) it is fun to overclock. kinda like people who work on cars. not necessarily especially productive, but we do it anyway
b) some CPUs do produce noticeable gains. Duron 600's are known for their ability to OC to up to 1ghz, which when you are playing games, WILL help significantly in UT - it will be noticeably smoother. Of course, my tbird 700 doesn't do 800 on air, so OC'ing it is really not worth it unless I went with watercooling.
One clear advantage of water is noise. With a good pump, and a large, slow radiator fan, a water-cooled system can be silent.
Actually, ALL athlons INCLUDING the classic can do SMP. The problem was lack of motherboards/a chipset. A slot Athlon classic should work in SMP, as should a duron. For that matter, the EV6 bus DOES NOT require the "S" in SMP - you could run a duron + tbird if you want (someone did it). I dont think it performs well like that, though!
Where might the uninformed learn about what pipes are? I'd like to know what these are that suck so much on this OS I just installed and have no issues with.
You know I have a drive image that does almost what you need... I would have given it to you if you had asked. Anyway - it would be much cooler here in E because its a bigger bathroom;-)
you REALLY need to go do some reading. when did it begin, conspiracy theorist? was a 386 also half speed? IF SO... why is it possible to (on most x86 chips) to VERY ACCURATELY guess clock speed by just counting to 300 million in assembly (about 3 instructions per number - decrement, compare, jump) and time the execution? if my 700 really is a 350, then that would be REALLY impressive if the cpu could somehow take advantage of any parallelism in that code. here it is for the curious:
100 MOV ax, FFFF
102 MOV bx, FFFF
104 DEC bx
106 CMP bx,0
108 JNE 104
110 DEC ax
112 JNE 102
114 MOV ax,4c
116 INT 21
**** NOTES ****
1. the instruction numbers are probably wrong - i dont think every one of those is really two bytes
2. the two MOVs at the start use wrong values. figure out yourself what to use. If its not high enough, add a cx loop on the inside.
3. run it at your own risk. hopefully you're doing this on an x86 machine.
At work, along with some p3 600s, ppro 200s running windows (the ppros are SLOW), we have 2 85-mhz sparcs... they are pathetically slow. However, running everything remotely (netscape, etc) over a 10mbps network works great.... and the server is only about 400mhz. I imagine that extending this to more machines would work pretty well too.
what's wrong with using cheap PCs though? who says old isn't stable? for that matter - you've got it all wrong.... older stuff NEVER crashed due to poor quality hardware.
And why not just GET big monitors but use cheap machines? It raises the cost... but still, you're saving lots per machine. and you could throw them in fancy cases if the employees complain - then they REALLY WOULD NOT be able to tell the difference
Slashdot Mirror
Your information is old. SiS used to be low quality, but their recent SiS735 socketA chipset is one of the fastest available. When using motherboards based on the sis735 with onboard sound/NIC, it is easy to build a FAST and CHEAP and STABLE system. I can confirm this because my room-mate has one. It doesn't crash (winXP). It is fast. It was cheap.
I just helped a kid put together a $450 system based on the 735, and it performs very well too.
I don't know it would take that much work. I imagine throttling would be hard, but is it such a large step to go from being able to detect overheat to powering off itself like the P3 does?
If you're going to ask someone to risk their athlon, please demonstrate your pentium4. Specifically, I would be fascinated to watch a heatsinkless pentium4 boot into windows. That would be awesome. So... show me ;)
Yes, on the P3s, the cpu powers itself off, without needing any motherboard stuff.
Agreed. In Tom's video, the instant the heatsink came off, the thermal paste began to smoke and the system died. here, there was no visible smoke(or paste), and the system continued for a significant chunk of a second (maybe about 1/10th of a second or so). Something is severely wrong with one of the tests.
Even though it is possible, I would think Tom should have retested different CPUs and different mobos. You CANNOT generalize based on a sample size of one.
not exactly. in this case, intel would have to have the CPU realize, "oh, this is a standard benchmark i'm running. let me work at half the floating point accuracy so i can outperform the athlon"
the one thing I dont like about that is
. Only subscribers can submit stories.
I think that the priviledge to submit should be given to everyone.
thats what I did.
I didn't really like mandrake when i played around with it... does the default install not install binutils? i was trying to get the latest kernel installed, and it was a real pain. maybe i just dont do well with gui everything ;)
the disadvantage (possibly) is that it does not leave room for northbride improvement...
What is the max delta T across these devices? If you want to create a fanless system using one of these, the heatsink on the hot side has to disipate LOTS of heat. that means it will get a lot hotter than room temp, so if these devices can only create a 30C difference, you might only get slightly below room temp on the cold side
I recall reading something on the web (was it here?) where they timed some linux compiles, and the dual tbird was more than twice the speed of the single.
a) it is fun to overclock. kinda like people who work on cars. not necessarily especially productive, but we do it anyway
b) some CPUs do produce noticeable gains. Duron 600's are known for their ability to OC to up to 1ghz, which when you are playing games, WILL help significantly in UT - it will be noticeably smoother. Of course, my tbird 700 doesn't do 800 on air, so OC'ing it is really not worth it unless I went with watercooling.
One clear advantage of water is noise. With a good pump, and a large, slow radiator fan, a water-cooled system can be silent.
if anyone would care to explain / provide links that would be great
Actually, ALL athlons INCLUDING the classic can do SMP. The problem was lack of motherboards/a chipset. A slot Athlon classic should work in SMP, as should a duron. For that matter, the EV6 bus DOES NOT require the "S" in SMP - you could run a duron + tbird if you want (someone did it). I dont think it performs well like that, though!
Where might the uninformed learn about what pipes are? I'd like to know what these are that suck so much on this OS I just installed and have no issues with.
:-( i agree completely... the ads sucked
I agree with you completely, but I DAMN WELL BETTER NOT be paying for your healthcare through my taxes/bills.
You know I have a drive image that does almost what you need... I would have given it to you if you had asked. Anyway - it would be much cooler here in E because its a bigger bathroom ;-)
you REALLY need to go do some reading. when did it begin, conspiracy theorist? was a 386 also half speed? IF SO... why is it possible to (on most x86 chips) to VERY ACCURATELY guess clock speed by just counting to 300 million in assembly (about 3 instructions per number - decrement, compare, jump) and time the execution? if my 700 really is a 350, then that would be REALLY impressive if the cpu could somehow take advantage of any parallelism in that code. here it is for the curious:
100 MOV ax, FFFF
102 MOV bx, FFFF
104 DEC bx
106 CMP bx,0
108 JNE 104
110 DEC ax
112 JNE 102
114 MOV ax,4c
116 INT 21
**** NOTES ****
1. the instruction numbers are probably wrong - i dont think every one of those is really two bytes
2. the two MOVs at the start use wrong values. figure out yourself what to use. If its not high enough, add a cx loop on the inside.
3. run it at your own risk. hopefully you're doing this on an x86 machine.
you misinterpreted that. that is just to keep the math right so everything works out nicely
sciam had a blurb that mentioned a yet unpublished study on purring and bone growth
At work, along with some p3 600s, ppro 200s running windows (the ppros are SLOW), we have 2 85-mhz sparcs... they are pathetically slow. However, running everything remotely (netscape, etc) over a 10mbps network works great.... and the server is only about 400mhz. I imagine that extending this to more machines would work pretty well too.
what's wrong with using cheap PCs though? who says old isn't stable? for that matter - you've got it all wrong.... older stuff NEVER crashed due to poor quality hardware.
And why not just GET big monitors but use cheap machines? It raises the cost... but still, you're saving lots per machine. and you could throw them in fancy cases if the employees complain - then they REALLY WOULD NOT be able to tell the difference