Author: Robert Hyatt
Date: 08:34:59 04/10/03
Go up one level in this thread
On April 10, 2003 at 02:37:36, Aaron Gordon wrote: >On April 09, 2003 at 23:08:15, Robert Hyatt wrote: > >>On April 09, 2003 at 20:17:14, Aaron Gordon wrote: >> >>>On April 09, 2003 at 18:01:50, Robert Hyatt wrote: >>> >>>>On April 09, 2003 at 15:36:58, Aaron Gordon wrote: >>>> >>>>>On April 09, 2003 at 12:14:37, Robert Hyatt wrote: >>>>> >>>>>>On April 09, 2003 at 11:04:51, Aaron Gordon wrote: >>>>>> >>>>>>>On April 09, 2003 at 09:24:01, Charles Worthington wrote: >>>>>>> >>>>>>>>On April 09, 2003 at 00:46:15, Charles Worthington wrote: >>>>>>>> >>>>>>>>>On April 09, 2003 at 00:34:10, Pavel Blokhine wrote: >>>>>>>>> >>>>>>>>>>On April 09, 2003 at 00:17:16, Charles Worthington wrote: >>>>>>>>>> >>>>>>>>>>>On April 08, 2003 at 23:53:06, Charles Worthington wrote: >>>>>>>>>>> >>>>>>>>>>>>On April 08, 2003 at 23:41:44, Pavel Blokhine wrote: >>>>>>>>>>>> >>>>>>>>>>>>>I am looking to buy a new computer. I will be using it for many things, but >>>>>>>>>>>>>mostly for computer video games and chess online. So what computer would be >>>>>>>>>>>>>best? TheMicro Express MicroFlex 27A Powered by AMD's 2.17-GHz Athlon XP 2700+ >>>>>>>>>>>>>processor and 512MB of DDR400 SDRAM, the MicroFlex 27A earned a blistering score >>>>>>>>>>>>>of 130 on PC WorldBench 4 tests, for $2254 or a Dell Dimension 8250• 3.06-GHz >>>>>>>>>>>>>Pentium 4 processor• 200GB hard drive• 18-inch LCD monitor• 128MB ATI Radeon >>>>>>>>>>>>>9700 Pro graphics• 16X DVD-ROM drive, DVD+RW/+R drive at $3158? >>>>>>>>>>>> >>>>>>>>>>>> >>>>>>>>>>>>If video games and chess are your main uses for the machine I would think the >>>>>>>>>>>>AMD would be the better choice...especially for the money....I own a dimension >>>>>>>>>>>>8250 and while it surely performs well it was hardly worth the additional cost >>>>>>>>>>>>were i to only use it for chess. But in the end it really is just a question of >>>>>>>>>>>>personal preference. If both companies offer comparable service and support then >>>>>>>>>>>>its really a toss of the coin. >>>>>>>>>>>>Charles >>>>>>>>>>> >>>>>>>>>>> >>>>>>>>>>>P.S. The decision will also depend on what you plan for the machine in the >>>>>>>>>>>future. If you want to upgrade to a faster cpu later on or overclock your >>>>>>>>>>>current one then the AMD is the only option for you. Dell boards do not support >>>>>>>>>>>overclocking and Intel is notorious for often changing the socket on their cpus. >>>>>>>>>>>So you can pretty much rule out upgrading the Dell at all without >>>>>>>>>>>some_major_expense. Speed wise you are not going to see a huge world of >>>>>>>>>>>differece between the two machines. Assuming no overclocking you will get close >>>>>>>>>>>to 1200 kNs running fritz 8 on either. The hyperthreading will give you a 10 to >>>>>>>>>>>12% boost with the intel running Deep Fritz 7 but the strength difference is >>>>>>>>>>>negligible. >>>>>>>>>>> >>>>>>>>>>>Charles >>>>>>>>>> >>>>>>>>>> >>>>>>>>>>Thanks. But how do i overclock an AMD and is it safe? >>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>>>Well, I am not an expert in overclocking but for mild overclocking, where no >>>>>>>>>additional cooling is required, i think you can just bump the cpu clock speed up >>>>>>>>>in bios say 10% or so safely (my figure may be inaccuarate). You may have >>>>>>>>>configure a jumper or two on the system board as well I am not certain about the >>>>>>>>>design of these boards. Aaron Gordon would be the one to pose this question to >>>>>>>>>as he has extensive knowledge of these boards and overclocking techniques. Sorry >>>>>>>>>I couldn't be more help. >>>>>>>>> >>>>>>>>>Charles >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>>The same also applies to the dual 3.06 xeon system you emailed me the specs to >>>>>>>>as we have already discussed above. The xeon systems have decreased in price a >>>>>>>>bit though. For chess I would recommend a minimum of 1GB of RAM and 2 would be >>>>>>>>even better but like all things related to the computer it is a matter of what >>>>>>>>you can afford to sink into one. Dell's business machines are more flexible as >>>>>>>>far as upgradability than their residential machines however they still cannot >>>>>>>>be clocked up. My machine is automatically upgraded each time a newer or faster >>>>>>>>part comes out as per my contract with Dell. The day after the part is released >>>>>>>>a technician comes to install it. This contract includes motherboards, cpu's, >>>>>>>>and memory only. As far as overclocking goes there is little need to overclock a >>>>>>>>high end AMD or Xeon dual. They are both quite fast as it is and with the upper >>>>>>>>end cpu's if you try anything more than mild overclocking you will need >>>>>>>>additional cooling because these cpu's are already clocked close to the upper >>>>>>>>end of what the engineers say is the limit of the part. Taking a 2 GHZ part to >>>>>>>>2.5 GHz is much simpler than taking a 3GHz part to 3.5GHz. Like I said earlier >>>>>>>>though, these are questions best left to Aaron Gordon or one of the guys here >>>>>>>>who specialize in overclocking. Personally, with the current advances in cpu >>>>>>>>technology i think overclocking the high end parts is a bit silly when there are >>>>>>>>faster and faster cpu's coming to market constantly. Why take the risk when it >>>>>>>>isn't needed? Most overclocking is done on the lower end chips to bring them up >>>>>>>>to where the higher end chips are without putting out the expense for the high >>>>>>>>end one. I think Aaron would agree that overclocking a high end chip past it's >>>>>>>>safety margin would be unwise. >>>>>>>> >>>>>>>>Sincerely, Charles >>>>>>> >>>>>>>Overclocking a high-end chip is fine as long as you know what you're doing. Also >>>>>>>as long as you run stability tests to ensure whatever cpu you're overclocking is >>>>>>>completely stable it doesn't matter if you're overclocking an AthlonXP 3000+ or >>>>>>>a 386SX-16MHz. Stable is stable. >>>>>> >>>>>> >>>>>>_IF_ it is verified. This is not easy. It means you have to run a program that >>>>>>specifically >>>>>>tests "edge conditions" by running sliding 1 patterns thru every instruction, to >>>>>>see if there is >>>>>>any unexpected cross-talk at higher frequencies, etc. >>>>>> >>>>>>Just because it runs some application correctly, does _not_ mean it will run >>>>>>_all_ applications >>>>>>correctly. >>>>> >>>>>With all of the tests on my box that I've done I'd be willing to put money down >>>>>that it is 100% stable. I'll put up a linux shell to that particular box and you >>>>>can try everything in your power to crash it via cpu/memory/chipset/etc >>>>>instabilities, it's just not going to happen. >>>> >>>> >>>>You could be right. But then, does the _average_ overclocker do that much >>>>testing? >>>> >>>>Nope... >>> >>> >>>Well, every overclocker I know runs either burnk7 or prime95. Both of which will >>>knock your box down QUICK if there are any instabilities. There are sets of >>>tests people run. Usually it's Prime95/Burnk7 for the CPU, memtest86 for the >>>memory and hours of 3DMark2001SE for videocard testing (when overclocking the >>>videocard or AGP bus). >>> >>>Current Nforce2 boards allow you to lock the PCI and AGP at default speeds so >>>you don't have to worry about pushing your videocard and PCI devices beyond >>>specs. Thats something I really don't like doing, but have done it in the past >>>because it was necessarry to get higher bus speeds. >>> >>>Most people just going from say 2400+ to 2600+ really won't need much testing, >>>the chips can do a little bit without hardly any risk of instabilities. It's >>>just when you start pushing far above the manufacturers top cpu is when you need >>>the extreme testing. Mainly hardcore overclockers do this and they run the same >>>tests I do. >> >> >>I'm not going to comment on those programs. However, they very likely do _not_ >>represent the worst case for catching problems such as slightly long gate >>delays, or cross-coupling on traces, or instabilities at certain clock speeds. >> >>It's non-trivial and requires specific testing rigors. A prime number tester >>is serendipitous at best. It almost certainly doesn't execute all instructions, >>which is a problem... Not to mention all instructions with specific "tough" >>bit patterns (many adjacent 1 bits, many adjacent 0 bits, sliding 0101/etc >>patterns... etc.) > >Knock it all you want, there's no way you'll crash my box though. ;) I'm not knocking your box at all, as I said previously you have driven it to the lowest latency I have ever seen. But for every success story, there are a thousand horror stories. I've seen too many of 'em here. "It works fine on all the normal test programs but I am getting a program crash on my real application. When I turn the clock back down, it works fine." That's not the way to spend time debugging...
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.