| |||||||
| Linux OS Problems General Linux-based OS problems. |
![]() |
| | LinkBack | Thread Tools | Display Modes |
| | #1 (permalink) | |
| So I'm digging through some old hardware, trying to see what's worth having. One item is an ECS L4VXA2 rev. 1.0 board with a P4 2.66 on it. Since it's the only P4 board I have handy, I can't do much swap testing. Another item was a 512M stick of PC2100. The two should go together well, with the CPU's 533MHz bus speed. So obvious thing to do: Bolt it together and run Memtest before going to huge efforts trying to build a whole box. So I've got it sitting in a spare case, with my fancy-ass server PSU driving it, and memtest86 3.2 running. It's working okay for now... BUT it reports weirdly: Pentium 4(0.13) 2659MHz L1 Cache 8 21793MB/s L2 Cache 512k 18592MB/s Memory: 512M 692Mb/s The 692Mb/s seems very low even for DDR2100... wasn't the whole significance of "2100" that it was somewhere in the range of 2100MBps peak throughput? The '8' for cache is also odd perhaps. It says "RsvdMem: 20M" -- I seem to recall this being like 100k on other systems. Odd? How is the draw of one of these boards/CPUs with three sticks of memory gonna compare with a Palomino-core 2000+ on a KT333 board wioth two sticks? I note the 2000+ is reported as a 70-watt CPU (per the listings on hwbot.org) and the 2.66B is listed at 68 or so, but I suspect the P4 style power distribution is different from the K7 board's (and the old but extremely solid PSU I bought back when an Athlon 1200C was a hot CPU and the P4 was brand new, if not earlier. | ||
| | | |
| Sponsored Links |
| | #2 (permalink) | |
| Couple of things here... First..it sounds like since you have just ONE stick of memory, then you bandwidth may be lower than you might expect since it would only be running single channel! Second, the timings of the ram are most likely default probably not "Optimal" for gettting good numbers, especially if it is not "Performance" ram. Using 3 sticks of ram is usually not a good idea, unless it is a "Single Channel" board to begin with, and then more ram won't effect the bandwidth to any degree. I would use 2 or 4 sticks, 1 and 3 are generally a waste on a dual channel board...there were some rare execptions with NF-2 where you could use a 512 stick and 2x256 sticks in specific slots and still keep dual channel. Have you looked at CPUZ to see what timings it's actually using and other info you can find out? How about trying Si-Soft Sandra 2004 or 2005 (Skip using 2007) or Everest to verify the bandwidth? Last edited by $SOLID$ Necro; 07-January-07 at 02:07 AM. | ||
| | | |
| | #3 (permalink) | |
| Well, it was just running as the board with one stick of memory (yes, single channel) in the case, because I didn't want to muddle through the endeavour of finding a hard disc (there's a 5Gb one *somewhere* in my house, but where?) and installing an OS on it until I was confident it was functioning in a marginally correct manner. | ||
| | | |
| | #4 (permalink) | |
| Yeah..running me test without a OS and just one stick of ram is probably not a good way to judge how it's "Real" performance will be, since there are "Optimizations" that are going to happen when you run Windows XP...remember that is part of the formula needed when AMD started using the "XP" tag on it's CPU's! (You didn't get the "Full" performance from them on DOS, Linux,Win 95, 98 or ME) | ||
| | | |
| | #5 (permalink) | |
| :surprised look: I've never heard that before... although honestly EVERY CPU I've tried has performed anemically under 2000. -- it seems like boot times are basically the same for a Celeron 533/400M, a Duron 1600/256M and a Athlon 4600+/1G. | ||
| | | |
| | #6 (permalink) | |
| Your boot times are not drastically going to be different no matter what ram (As long as its 1GB or more) or CPU speed you use! The two biggest factors are typically your OS and your HD! I had a 36GB raptor, and the damn thing would boot before the second round of the "Bar" that you see as XP loads! With a 80GB WD IDE, it usually takes 3 or 4 spins before windows goes to the desktop. I know of plenty of friends rigs that take as many as 10 spins or more to get to the desktop with all the crap they have loading, and not to mention they never do a disk clean up and defrag on a regular basis. And as far as your OS goes, there are "Tweeks" that can make it load faster (There are plenty of optimization guides on the net if you google them) not having a ton of things that start up automatically will certainly be noticable..Anti Virus and ATI's"Control Panel"' are two of the biggest "Bogs" to a fast booting rig. | ||
| | | |
![]() |
| Thread Tools | |
| Display Modes | |
| |
Similar Threads | ||||
| Thread | Thread Starter | Forum | Replies | Last Post |
| Fails Memtest86+ But Does well Elsewhere? | jedihobbit | Memory (RAM) OC | 1 | 09-February-07 12:26 PM |
| Bandwidth Use | TheGreatSatan | Coding / Scripting / Programming | 6 | 10-August-05 08:31 AM |
| Bandwidth limiter | Crazy_J | Internet / Network Tweaks | 4 | 30-November-04 11:08 PM |
| Errors in memtest86 with ocz el ddr | emopunker07 | Other Hardware | 17 | 17-April-03 12:38 AM |
| Bandwidth | Darksamurai | Daily Disturbance | 4 | 31-January-03 03:48 AM |