Mushkin's Friday night Special PC-150 for $99/128MB DIMM

[xortam]

I found out some of the info in the ...\Docs\common\qa\det_mem.htm page of the on-line documentaion.

What does the PCx-xxx-xxx marking displayed by Sandra mean?
A: There are many conventions:

PC66/100 SDRAM Intel Specification - Version 1.0 to 1.2:
PCx-abc-defm (e.g. PC100-322-620R) where:
x - Speed rating (MHz). Memory bus speed should be equal or lower than this.
a - CAS latency (cycles). Lower the better (faster), but more expensive.
b - RAS to CAS delay (tRCD cycles). Lower the better.
c - Row precharge (tRP cycles). Lower the better.
d - Read data access time (tAC ns). Lower the better.
e - SPD chip revision. 2 corresponds to SPD 1.2.
f - Design revision. 2 corresponds to Rev 1.2.
m - R for registered DIMMS. 256MB and bigger modules need to be registered.
PC66/100 SDRAM Extended Intel Specification - Version 1.2+:
PCx-abc-ddeefm (e.g. PC100-322-54120R) where:
x - Speed rating (MHz). Memory bus speed should be equal or lower than this.
a - CAS latency (cycles). Lower the better (faster), but more expensive.
b - RAS to CAS delay (tRCD cycles). Lower the better.
c - Row precharge (tRP cycles). Lower the better.
dd - Read data access time (tAC ns). 54 corresponds to 5.4ns. Lower the better.
ee - SPD chip revision. 12 corresponds to SPD 1.2.
f - Design revision. The current is 2 corresponding to Rev 1.2.
m - R for registered DIMMS. 256MB and bigger modules need to be registered.

Looks like my memory is running at CAS 2, RAS to CAS 2 (tRCD), and precharge 2 (tRP) (PC100-222-620) and 6ns read (tAC), SPD 1.2, rev 0 (PC100-222-620). Your 452 number means that you have 4.5ns memory. Still haven't found an explanation of my "7-1-1-1R 5-1-1-1W" numbers.

[ALBPM]

Thanks Xortam

Paul

[xortam]

Well I just got around to installing my new Mushkin PC-150 memory (installation delayed a few days for various reasons). Seems it doesn't like working with my old memory (see specs above). I moved my old memory to slot 3, installed the Mushkin into slots 1 and 2, and the system would boot up but hang while bringing up the desktop (first time got up to inspecting the device manager). I removed my old memory and everything appears stable so far. I'm glad I landed getting that second stick because that's the only way I'd be running with 256MB now. I haven't tried shuffling the memory around more but I will try later in the day. The BIOS now reports 3-2-2 (SPD) settings where I was getting 2-2-2 (SPD) before with my old memory (shouldn't it be 2-3-2, note 100 Mhz FSB ). Sandra now reports: Bank 0 + 1 and 2 + 3 Memory Modules: 128MB 16x(8Mx8) SDRAM CAS3 PC133U-333-542, Bank 0, 1, 2, and 3 setting: 64MB SDRAM 8-1-1-1R 6-1-1-1W. Odd that it reports -333. I could also try to force the memory timings in my BIOS but there were only two profiles available (can't recall details). Might try newer MB BIOS. Any suggestions to get these to work together? Or I could skip it and go onto trying to overclock with the new memory and ignore the old stick.

------------------

• ASUS P2B-S Rev 1.03, PIII 450MHz, Award ACPI BIOS v1010, 128 MB RAM
• MYLEX FlashPoint RAID+ (BIOS v2.02N) running RAID 0 on two 9 GB IBM DDRS 39130D Disks
• Diamond MX300 sound card, now with MX25 S/PDIF output
• Matrox Millennium G400 Max Dual Head - English
• NEC 5FG monitor
• Logitech MouseMan Wheel
• YAMAHA CRW4416S and NEC Multispin 3x CDs
• 3Com Fast EtherLink XL 10/100Mb TX NIC (3C905B-TX)
• US Robotics 56K Voice FaxModem Pro
• Pioneer DVD-303S SCSI
• Note--All SCSI devices (except disk drives on RAID) are connected to onboard AIC7890 U2W SCSI
• Mainly running Win98 v4.10.1998

[ALBPM]

Set it at CAS=2, CAS to RAS=3 and Precharge time=2

Paul

[xortam]

Gee, thanks for the obvious Paul.
Mushkin spec'ed the DIMM to run at 2-3-2 and I just assumed that's what the SPD would read but my BIOS set it at 3-2-2. Either there's some problem with my MB/BIOS and this memory or the DIMM EEPROM is not programmed as expected. Going back into my BIOS, I found that I do have complete control of the timings beyond the couple profiles provided. I set the timings at 2-2-2 (since my old PC100 RAM ran that at 100MHz) and the system seems to be stable. Sandra benchmarked the memory a little faster:
3-2-2 gave me CPU 248 MB/s, FPU 226 MB/s
2-2-2 gave me CPU 260 MB/s, FPU 237 MB/s (first run)
---------- and CPU 263 MB/s, FPU 244 MB/s (second run)

Sandra also now reports 7-1-1-1R and 5-1-1-1W just as it did with my old PC100 2-2-2 memory. Apparently those numbers (I assume read and write) are directly related to the memory timing parameters. It also still reports 128MB 16x(8Mx8) SDRAM CAS3 PC133U-333-542 which indicates 3-3-3 timings which are again different from what was detected through SPD. How is Sandra obtaining this information? Are the 2-3-2 timings just what Mushkin has tested this DIMM at when running 150MHz and nothing to do with the SPD programming?

I guess I'll try adding back my old memory and try running with these forced settings. Don't know why it would have had problems running at a slower rate along with the Mushkins.

[Greebe]

xortam, what cpu,/fsb speed are you running to get those benchmarks?

[xortam]

The benchmark is from SiSoft Sandra Standard v99.8.5.30 running on Win98 on 100Mhz FSB PIII-450 (system specs are posted above). Those numbers compare against reference numbers for a PIII-500 128MB system getting CPU 280 MB/s and FPU 260 MB/s.

[Greebe]

I was wondering if you were running the same cpu as those scores didn't look like a PC133 speed scores, hence my question. I own Sandra Pro ver, 2000.7.6.49 and wondered if there was a difference between them.