I-Appliance BBS
The Official Source for Internet Appliance Upgrades and Mods
Amazon Honor System Click Here to Pay Learn More
BBS Main List | Sign In | Sign Up | Search | Help | Linux-Hacker.netReply to Thread | Printer |

Home / I-Opener Areas / I-Opener Technical Stuff
Memory Interleave Revisited
Interleave enabler

New MessageMemory Interleave Revisited (modified 0 times) ckbone
Profile
Feeling lucky? Try this 4-way Memory Interleave program.....It's an easy install and you can uninstall it in the control panel. It works in Win98, ME, and 2k. Be aware this program could cause all sorts of damage....but it installed without a problem for me. The particular type of memory chip you have may produce different results. I tried it on a V4b with a K62-400, a Kingston 128meg pc100 chip, Win98. Too bad for me....The sandra benchmarks showed no improvment, but I'll try it on my other two I-Openers (different chips). Be sure to read the documentation. I could never get this page to load using Netscape....I used I.E. Scroll to the bottom of the page....It's a free program written by George Breese...Version 0.12. Please post any results....

http://www.viahardware.com/download/viatweak.shtm

06-13-2001 17:47:12

New MessageRE:Memory Interleave Revisited (modified 0 times) ckbone
Profile
I installed this prog on a V2, winchip2, PNY 128meg PC100. Again with no problems....This time the Sandra memory benchmarks improved 20% with no other changes, proving to me the program helped. The system does feel faster. Unfortunately there is no other way to prove that the program was of any benefit. Perhaps a program that is memory intensive, like photoshop, could be timed on a specific operation. Anyway, I'm impressed.
06-14-2001 03:16:20

New MessageRE:Memory Interleave Revisited (modified 1 times) Programmer
Profile
There's actually a memory interleave enabler written by someone here on the board... ME!! in any case there's a couple versions, including one that can be inserted into the bios(for those using ME/NT/2000/linux/etc), and one that can be run from the autoexec.bat. They're REALLY small and should download in less than 1 sec on a 2400bps modem..

The standalone dos version (should be run from autoexec.bat)
http://bethie.net/~programmer/SetInter4.com

The bios insertable version(use cbrom) also does FixUSB and sets video mode to 3 (color text 80)
http://bethie.net/~programmer/bi4.bin

06-14-2001 07:32:30

New MessageRE:Memory Interleave Revisited (modified 0 times) Programmer
Profile
A Note: I looked at George Breese' page, and:
1) the functional part of the memory interleave is the same(there's only one way to do it)
2) FixUSB is similar to his PCI latency software but I change the latency timer on offset 0D(where the chipset doc says to) and he apparantly changes it (and other stuff) on offset 75. I also twiddle some stuff in offset 70(which he does not do)

3) PRIMARY DIFFERENCE: he makes his changes using windows drivers which occupy memory and can be removed with add/remove programs. I make my changes either in dos with no resident code, or in the bios with no resident code. mine are also somewhat smaller.. WAY under a k each..

06-14-2001 08:06:07

New MessageRE:Memory Interleave Revisited (modified 0 times) ckbone
Profile
Programmer.....I used George's program instead of yours only because of feedback. The VIA Hardware forum has 38,000 members vs 2200 or so here. They have a whole section devoted to Georges patches and VIA tweaking in general. If you have any interest in VIA related problems and fixing, that is the place to go. When the new "fixed" BIOS chip with your patches is available, I'll be the first in line to try it. Till then, I read both forums....try to figure out the best course of action from all available information....your info has been extremely helpful, probably the best on this board. I wish more people here would comment on your fixes. Feedback is what drives this BBS.
06-14-2001 09:50:27

New MessageRE:Memory Interleave Revisited (modified 0 times) Wild_Pencil
Profile | Email
ckbone,

I think you misunderstood Programmer's post. His BIOS upgrade is available *now*:
http://www.bethie.net/~programmer/bi4.bin

Use cbrom to insert it into a copy of your BIOS:
cbrom v500.bin /isa bi4.bin

At present, this is the most painless way to get the "FixUSB", "Memory-Interleave", and "Mode CO80" fixes onto your IOpener.

The BIOS I'm working on will have a much nicer way to invoke the Smart BootManager when the IOpener boots up -- using "B" as a hot-key during the bootup sequence.

I'm also fiddling with the Setup-Screen, and have the Floppy-Drive partially enabled. I still need to add a 3rd-Party "Int13" handler to let the BIOS actually USE the floppy.. The version from the "TinyBIOS" project needs to be reformatted for NASM, and I'm not quite up to speed on the ASM format TinyBIOS's author used.

All in all, this summer will be quite interesting for IOpeners when the firmware upgrades are done.


I definitely have to thank Programmer and Turbo3 for all their help in getting things this far. Almost all of the hard work has been done by these two.

-WP

06-14-2001 10:32:14

New MessageRE:Memory Interleave Revisited (modified 0 times) ckbone
Profile
Wild_Pencil.....My 3 I-Openers have early Badflash BIOS chips, not flashable in the I-Opener. I'll have to buy a new chip to get the new features.....
06-14-2001 13:34:29

New MessageRE:Memory Interleave Revisited (modified 0 times) ckbone
Profile
Wild_Pencil.....My 3 I-Openers have early Badflash BIOS chips, not flashable in the I-Opener. I'll have to buy a new chip to get the new features.....
06-14-2001 13:34:59

New MessageRE:Memory Interleave Revisited (modified 0 times) *SF*
Profile
Dude, Don't You have the original BIOS Chips ? (HotFlahing could Work. It did For Me.)
Nothing Wrong with Choosing http://www.BadFlash.com He Provides a Great Service & is Safer !
I would Recommend anyone use a Real Chip Extractor.
I Think Most of the BIOS Contact Hassles were caused by UnEven Pulls
(Yea, the Epoxy didn't Help 4,000 Units Either !)
The Chip Extractor does a Nice Straight Pull.
worth the $10 @ Radio Shack Not to Have a Bent Contact.

for Neophytes: "HotFlashing" is Pulling the GOOD BIOS Chip out of a Running I-Opener
to stick in a Non Harddrive Bootable BIOS Chip & Reflash it. (The Original V5 BIOS Chips)
This Carries a Certain Amount of RISK !
You Need to Have an Idea What You are Really Doing & not just be followig a Step by Step.
Also using a UPS is Recommended.

Once You Have a Bootable BIOS, You Don't Pull it Out. You Don't HotFlash, You just ReFlash.

06-14-2001 14:23:17

New MessageRE:Memory Interleave Revisited (modified 0 times) ckbone
Profile
Look...for the limited benefits involved here, it is not worth risking a fried computer....hot-flashing is not my idea of a risk worth taking. When Badflash has a new, fixed BIOS available, I'll buy one, or send him mine to re-flash. I'm waiting for a BIOS that will boot a cdrom, anything else is just icing. This memory interleave is a worth-while hack...try it till the new BIOS is available.
06-14-2001 16:28:52

New MessageRE:Memory Interleave Revisited (modified 0 times) Programmer
Profile
The memory interleave IS a worthwhile hack, and has been available for months, even for user of linux, qnx, and anything else. It is a tiny download indeed, and easy to set up on DOS, 95,98, and ME. The BIOS version is slightly more painful, and carries with it the risk (however minimal) of misflashing your bios. There are no drivers to consume resources, no memory resident software to fail, just configure the chips and go away..
06-15-2001 05:53:04

New MessageRE:Memory Interleave Revisited (modified 0 times) mp3boombox
Profile | Email
Its more then that, its like 3fold better!!

much faster access lots of things dont have mini pauses any more. I cant even beggin to count all that the interleave has done.

06-15-2001 18:31:23

New MessageRE:Memory Interleave Revisited (modified 0 times) zyxw
Profile | Email
Sigh,
Ok... help me out here... CBROM is needed prior to flash?

A flash utility I have, I'm unfamiliar with CBROM, where is it, and can someone verify the step by step (also why wasn't a patched version just posted to allow for easy flashing without an intermediate step? Does CBROM permit successive additions to the bios or some such??)?

Thanks

06-15-2001 20:34:25

New MessageRE:Memory Interleave Revisited (modified 0 times) Programmer
Profile
No GOOD reason.. just 256K is 2405 times larger a download than 109B.. I guess I can live.

Version 5 Patched + BI4.bin: http://bethie.net/~programmer/v5pMI.bin

06-16-2001 06:04:13

New MessageRE:Memory Interleave Revisited (modified 0 times) Programmer
Profile
also.. if you have some OTHER bios, you still probably want to go the cbrom route.. (Wild_Pencils cd bootable for instance)
06-16-2001 06:06:44

New MessageRE:Memory Interleave Revisited (modified 0 times) ckbone
Profile
I've tried George Breese's program on three I-Openers and a Compaq desktop (it has the same VIA chipset as an I-Opener). Two of three I-Openers showed no improvement...the Compaq showed an FPU bandwidth imrovement of 35%...quite a big step. I think the I-Opener improvement can be traced to the specific memory chip....in this case a PNY 128meg chip I bought from BestBuy. If anyone here gets benefit from any Interleave program, please post the chip brand, or any other useful info.
06-16-2001 07:01:39

New MessageRE:Memory Interleave Revisited (modified 0 times) mp3boombox
Profile | Email
dont quote me but i think the crucial memory has the best preformance ratting????

I'v got the ct16m6486w8e.8t memory??

128mb 144-pin sodimm 16Mx64 sdram pc100

My HD is a ??? brand 12gig hard drive. 9mm with udma 33, my hard drive access like any laptop drive is slower then death especialy when loading something but hay thats no biggie. I do wish i could get the 3d part of the chipset working though, would be realy nice to use my io as a portable gaming system. as is i run counter strike on it just fine in software at lowest res and like no effects. But it play's nicely. If the 3D ruteans could be setup then the thing would be like a small power house.

I have a friend with a trident blade3D video card and its not the greatest but it DOES DO THE JOB.

and the cyberblade i HAVE confermed is 3d compatable, its based on the blade 3d chipset but the call ruteans have not be implimented period in the drivers.

I'm trying to get a friend to try and make a basic driver to enable a few call's like blur filtering and zbuffer. that alone would make a BIG improvement.

06-16-2001 20:28:14

New MessageRE:Memory Interleave Revisited (modified 0 times) skitch
Profile
what do you use to flash the .bin into a bios?
06-17-2001 14:15:44

New MessageRE:Memory Interleave Revisited (modified 0 times) Programmer
Profile
If you've downloaded http://bethie.net/~programmer/v5pMI.bin download it again and reflash.

If you've used the method posted by Wild_Pencil to add bi4.bin to your bios, you'll need to get a clean BIOS and redo the cbrom. The method listed does not seem to work. The bios loads the image into somewhere where it does not get recognized.

This is the method that puts it in the right place;
cbrom <BIOS Name> /other 8000:0000 <ROM Name>
i.e. cbrom v5pmi.bin /other 8000:0000 bi4.bin

Please make sure that you are using cbrom v1.30 http://bethie.net/~programmer/cbrom.exe

I've always used awardflash to flash my bios' http://bethie.net/~programmer/FL.exe

06-18-2001 06:18:16

New MessageRE:Memory Interleave Revisited (modified 0 times) Guy2000
Profile
I just flashed my V3 using the v5PMI.bin file and it works great. Thanks!

A dumb question: once I have a stable system I like to have a VGA boot logo, using my own bitmap to match my kitchen decor. I tried adding it in using CBROM but just get a blank screen for the duration of the POST... releasing it and the text screen works fine.

Should the V5PMI bios accept a boot logo? I presume the Mode CO80 switch may be confusing it.

Not a big deal as I can live with the text screen, but if you have a fully patched version 5 with FixUSB and Interleave, please let me know.

Also, how can I tell if the Interleave add-in is working? It would be nice to have a text message appear during the POST.

-Guy

06-18-2001 13:07:04

New MessageRE:Memory Interleave Revisited (modified 0 times) ckbone
Profile
The only way to tell if your memory interleave enabler has done any good is to benchmark before and after. Use the latest Sandra 2001 pro....current version. Here's the best benchmark of my three I-Openers (memory benchmark), after BIOS tweaking. PNY 128meg PC100-222-622....Winchip2 at 210mhz, Win98.

ALU....60
FPU....77

BIOS settings as follows:

three bank timings.....turbo
Sdram cycle length.....2
Dram read pipeline.....enabled
Sustained 3T write.....enabled
Cache R/CPU pipeline...enabled
cache timing...........fastest

The memory benchmarks increased almost 40% over the BIOS default settings and before I installed the Breese patch. I am really amazed how much difference these little tweaks, and the memory interleave has done to this system. A winchip2 shouldn't do this good.....I remember when I first tried to play an MP3 with this computer...bummer. Now it can handle almost anything I can throw at it.....including burning a cd over the USB. The memory scores of my other two I-Openers increased after the tweaks...but are still aprox 20% less than the PNY. Post your memory benchmarks and what you have done to achieve it....include the type of memory....maybe we can figure what works best on the I-Opener.

06-18-2001 18:38:38

New MessageRE:Memory Interleave Revisited (modified 0 times) Programmer
Profile
The inability to add a boot image is a problem with the bios v5_Patched. Wild_Pencil might be able to answer that for you. You can add the Option Rom to any bios you want, including your old V3 bios with the nice matching boot logo.. :) .. Enjoy..
06-19-2001 06:48:33

New MessageRE:Memory Interleave Revisited (modified 0 times) Wild_Pencil
Profile | Email
No idea why the V5_Patched wouldn't accept a new boot-logo.. though I do remember that there are some very broken versions of CBROM floating around out there. I use CBROM 2.07, which does recognize the "/isa" switch and put the extension ROM in the right place. Older versions, yeah.. you're better off using the "/other" switch as Programmer mentioned above.

Adding a boot-logo was always some sort of black magic. Take in a .BMP, convert to an EPA File, then flash it into the EPA region. To be safe, release the old EPA Logo first. My memory's a little fuzzy on this, but I think CBROM was written to embed multiple EPA logos.. even though our Award BIOS wouldn't know what to do with the extra sets...)

Have you tried running MODBIN against your BIOS Image after updating the EPA logo? In my setups, I always update the Version-String "(IPC_TRI) BlaBlaBla" before flashing, and I've never had a problem with BIOS Image files from MODBIN.

-WP

06-19-2001 09:27:28

New MessageRE:Memory Interleave Revisited (modified 0 times) Programmer
Profile
Guy2000: http://bethie.net/~programmer/bi4v.bin has a short message.. it also switches to mode 1 (40x25) then back to mode 3 (80x25). unfortunately, it increases the code size 50%, and I can tell you I'm not happy about that.. (it's like 150 bytes now.. :( ..)
06-20-2001 07:23:43

New MessageRE:Memory Interleave Revisited (modified 0 times) Guy2000
Profile
Thanks. I tried the new bi4v.bin

I released the old version, then used CBROM 2.08 with the ISA switch and it wouldn't take (not a valid bios file)

So I used the original /other 8000:0000 method and it inserted fine.

I reflashed, rebooted, and everything seems fine except that I don't get your nice message.

Perhaps I should just leave well enough alone, but the reason I'd like to see the message is to make sure that the Option ROM is loading properly on my system.

Any reason you can think why it won't display?

I haven't tried adding the boot logo yet, so am still in Text mode and everything else seems fine. I did a CBROM /D and the option is appearing in the list.

TIA

06-20-2001 12:49:04

New MessageRE:Memory Interleave Revisited (modified 0 times) Programmer
Profile
Guy2000: Try downloading and flashing http://bethie.net/~programmer/v5pmi.bin
it has the bi4v.bin option rom inserted with "cbrom /pci bi4v.bin" perhaps 8000:0000 isn't the right address.. (Though I thought it was)
06-21-2001 09:38:25

New MessageRE:Memory Interleave Revisited (modified 0 times) _wlj_
Profile | Email
I've got a v1 I-Opener w/ a WinChip 2 and 128MB PNY RAM running Windows 98. Interestingly, none of the software or BIOS tweaks described show any performance improvement under Sandra 2001's memory benchmark. I've tried the George Breese patch, as well as Programmer's v5pMI BIOS and bi4v.bin BIOS patch.

Here are the benchmarks I get using Sandra for both a stock v1 BIOS, as well as the various interleave tweaks:

Integer ALU/RAM Bandwidth: 41 MB/s
Floating Point FPU/RAM Bandwidth: 56 MB/s

Any thoughts as to why I'd see absolutely no change in performance?

Thanks!

06-21-2001 17:56:00

New MessageRE:Memory Interleave Revisited (modified 0 times) ckbone
Profile
Hmmm....You've changed the BIOS settings as I posted above...no improvement? Well, I can say this about memory chips: they all benchtest slightly different. I've tried four different chips on the same I-Opener (V2, Winchip2, Win98). The best two were PNY brand. However, one was made by Hyundai, the other Toshiba. The Toshiba was slightly faster. What I would say here is either there is a configuration problem with your OS, your BIOS settings are not configured properly for your specific chip, or the chip itself just wouldn't take this sort of tweaking. One way to check this is to buy another chip from Best Buy....see if it's faster with the same settings. You can always return it.
06-21-2001 18:31:13

New MessageRE:Memory Interleave Revisited (modified 0 times) _wlj_
Profile | Email
ckbone -

Yes, I've changed the BIOS settings to those that you had listed above. However, these more aggressive timings made no difference in the benchmarks. As far as configuration issues go, both the BIOS settings and re-flashing have been done correctly, so I don't believe there's any issue there. As to OS configuration, I've got a pretty vanilla installation of Windows 98SE. That, of course, doesn't mean that there aren't any problems with it as it is Windoze... ;)

I had made an error in my previous posting -- the memory module was not from PNY, rather from Simple Technologies. This 128MB Simple Technologies module has memory chips from IBM on it.

Based upon your suggestion, I "liberated" another 128MB SODIMM from my laptop for use in the I-Opener. This module is Toshiba-branded, with memory chips from Toshiba. Unfortunately, this new modules still yields the same 41 MB/s and 56 MB/s benchmarks.

I'm stumped as to why there's absolutely no difference between configuration settings, patches, and memory modules. I'm also disappointed to see that my box clocks in so far below the numbers that you've posted. I'd really love to be able to goose the performance of my I-Opener.

Thanks for any insights that you may have!

06-22-2001 00:18:57

New MessageRE:Memory Interleave Revisited (modified 0 times) ckbone
Profile
This whole tweaking thing is very hard to figure out, mostly because there are so many variables. For instance....I drop an image of my V2, Win98 build, onto a V4b, transfer the memory chip also, the BIOS settings are the same....you would think the memory benchmarks should be the same between the two I-Openers. They're not. The board version is different, the cpu is different, the hard drive is different. Why these differences affect the memory benchmarks, I can't explain. But something is going on here to change those memory benchmarks.

A word here about Benchmarks....In order to make any sense about them, you have to make certain that you have a very similar starting point before you run a benchmark. Make sure the temperature of the computer is at a similar point before running a benchmark. Very hot components will lower your scores. Turn off all TSRs before the testing, programs running in the background will skew your final scores. Only change one thing at a time...then benchmark. Multiple changes make any test scores difficult to diagnose.

In the case of those posted scores of mine, I can say this.....There has been no soldering on the M-board....none. The OS is Win98se, full version, clean install. A large number of tweaks have been done, using mostly free programs. The whole system is about 800mb, on a 1.4 gig IBM drive, dma enabled. It is a classic hack, mod2 kit, internal drive with a lasagna fan. I use the h.Oda FSB program to raise the the FSB to the "70" mark. This has the effect of raising the memory bandwidth to 105mhz....about a 5% increase. Without this tweak, the memory scores are 55/70. With the H-Oda program running, everything is faster...cpu scores, memory scores, and real world testing. In short, there is a virtually endless list of differences between my system and any other. Trying to explain why one system is faster, when most of the hardware components are the same, is very difficult. I do think a specific chip, brand or otherwise, is the biggest factor in attaining a higher benchmark. Usually a higher benchmark makes a faster computer. For me, I know that's a fact.

06-22-2001 04:00:51

New MessageRE:Memory Interleave Revisited (modified 0 times) Guy2000
Profile
Programmer, I tried the latest v5pmi.bin (the lower case version!) and reflashed. While my system continues to work fine, I am still not getting any startup message, nor any mode switching that I can see. I am not sure if I should be.

I tried releasing the /pci, and then reinserted the bi4v.bin that I had downloaded a couple of days ago (the one I was sure had the nice message) using CBROM208 and the /pci method. It reinserted fine of course, but still no message. While things are fine, I am guessing that I am not getting full advantage of the memory interleave code. I wonder if this is why _wlj_ is not seeing any performance change.

ckbone, are you using the BIOS patch or loading the fix after the OS loads?

By the way, this will sound strange, but one thing I find is this BIOS fixed a problem with a reset after screen power down that I have had in Win2k. Until I started using this BIOS, whenever I would reset the display (i.e., move the mouse) after it had dozed, the top 20% of the screen would not refresh correctly - it was black. What was even stranger was that this black band would appear on my other PC connecting to the Iopener via Remote Admin client (a VNC type application).... getting windows to redraw the screen (e.g., clicking on the Show Desktop shortcut) would restore the top part of the screen. Well with the V5 BIOS and the memory interleave/Fix USB patch, I don't get that problem anymore. The screen refreshes fine. I am using the latest Microsoft trident driver (the one with Windows 2000 SP2 and the newer one on Windows Update both had the same symptoms). Anyway, just thought I would mention it.

06-22-2001 06:48:56

New MessageRE:Memory Interleave Revisited (modified 0 times) Programmer
Profile
Guy2000 and everyone else: Eureka!!!

I found the problem.. The computer that I had tested howto make an option rom on (one I have at work) does not make a checksum of the rom. The IOpener bios does. The 8 bit checksum of an option rom must be 0. Hence I wrote a program (in VB) that takes a .com file and turns it into a .bin file that is an option rom and has a checksum of 0. http://bethie.net/~programmer/com2rom.zip

I also used it to create a new v5mpi.bin http://bethie.net/~programmer/v5pmi.bin

and a new bi4v.bin http://bethie.net/~programmer/bi4v.bin

06-22-2001 07:41:03

New MessageRE:Memory Interleave Revisited (modified 0 times) Wild_Pencil
Profile | Email
Programmer:

*bonk on the head* Ahhh... yeah, the checksum. Using the "/other" switch in CBROM doesn't enforce a checksum; while the "/isa" switch (in newer CBROMs) does enforce this before adding the ROM in.

06-22-2001 12:49:58

New MessageRE:Memory Interleave Revisited (modified 0 times) ckbone
Profile
Guy2000....I am not using any of Programmers patches. The George Breese program is a windows program, and loads its software automatically before the OS loads. No BIOS changes at all.....
06-22-2001 16:20:58

New MessageRE:Memory Interleave Revisited (modified 0 times) ratster
Profile
Programmer - Thanks again for the efforts. I'm still amazed you're putting up with all of my whining! You deserve a cookie for sure...

Regarding the latest BIOS image... Unfortunately, it didn't change anything. I snagged the latest v5pmi.bin file and flashed with it, but everything is still the same. Thinking that it would make a difference, I cbrom'd the bi4v.bin file to a copy of v5pmi.bin just in case and flashed that. No dice. As always, it definitely looks like the BIOS is being flashed (i.e. no errors beyond the "hey, this doesn't look like a BIOS image that you should be flashing" messages), so...

Sorry, I wish I had better news. This Award BIOS graphic boot screen is quite the pain in the arse, I must say... It's so pointless, and it's causing so much grief. Grrr...

06-22-2001 22:11:21

New MessageRE:Memory Interleave Revisited (modified 0 times) Guy2000
Profile
Yes, Programmer, I echo that. Make that two cookies. You're always so responsive with useful and elegant programs.

Anyway, I also tried the latest image and flashed it but with no apparent change.

I also tried CBROMing the image, first with pci and then with isa. The difference with isa this time round was that I did not get the "not a valid BIOS file" message, so things were hopeful. Feeling lucky I decided to MODBIN the ISA'd image with an updated version message (IPC_TRI) blah blah blah.

I reflashed.. reset... and I now have a dead iopener. I feel so silly... it was probably the MODBIN as I really had no idea what I was doing. Ive done a lot a flashing in my time, but never used MODBIN before, and now I know why. And ckbone, now I know why you went the windows driver route!

Anyway the machine turns on I can here the hard driver whirring.. but no display... and no booting into Windows.

I have a spare unhacked iopener in storage (just for such circumstances) so I'll have to pull that one out and hotflash the bios. Trouble is, its been a while so I can't even remember if mine was one of the gooped ones (I did the QNX flash routine from a hard disk), so I may have to get out the knife/hairdryer.

Any no worries, as I also have a BH6 lying around, which will save one less ungooping... but before I start all this if anyone knows of a way to recover without ungooping let me know. I am wondering if clearing the CMOS checksum or anything might help as when I changed the modbin message I didn't change the checksum or anything.

TIA
-Guy

06-23-2001 08:18:51

New MessageRE:Memory Interleave Revisited (modified 0 times) ckbone
Profile
I just spent a whole day trying to figure this memory tweaking thing. I'll try to explain my conclusions....I used three different chips, 128meg, Kingston, PNY/Hyundi, PNY/Toshiba. I used Win98 and Win ME, a V2, V4, V4b, a Winchip2, a Pentium200mmx, an AMD K62-400. For the Benchmarks I used Sandra 2001 pro.

It was immediately obvious the memory benchmarks were independent from the CPU and the OS and the I-Opener version.

George Breeses memory interleave program helped the score on only the PNY/Toshiba chip.

Using SoftFSB, at a setting of "70", raised the scores on all three chips.

Changing the BIOS settings helped all three chips....The PNY/Toshiba scored almost 50% greater than when I started with the following settings:

3 Bank timings.....medium
Sdram cycle lenght......2
Other settings posted above....These were the best settings with all three chips.

The Kingston chip did not respond much to any changes in the BIOS, SoftFSB helped it, however. Memory interleave..no.

The PNY/Hyundi was helped by BIOS changes, and SoftFSB. Memory interleave..no.

Here are the best scores, with those tweaks:

Kingston.....ALU 53...FPU 70
PNY/Hyundai....ALU 59...FPU 78
PNY/Toshiba....ALU 64...FPU 88

My conclusion at this point....It's the chip that makes all the difference, interleave works on some, not on others. Your individual BIOS settings have a big effect. Try changing the ones mentioned, benchmark each change. Your best settings with your own chip may be different from mine. SoftFSB helps all scores, memory benchmarks as well as CPU benchmarks. If various BIOS settings give you similar scores, go with the less aggressive setting, your system may be more stable.

FWIW....I did 10 hours of benchmarking on three different machines without one crash, proving to me the Breese interleave program is at least stable, helping some chips....not hurting the others. There are other BIOS settings that may help out, more tweaking to come........One more thing, I guess it benefits you to pay more for a better chip, ram prices being as low as they are. I want to see a benchmark using a Crucial/Micron chip.

06-23-2001 20:58:40

New MessageRE:Memory Interleave Revisited (modified 0 times) ckbone
Profile
I just found a neat program which gives detailed info about your ram chip....you might find it useful when tweaking memory settings.....Find the link to CTSPD here:

http://www.viahardware.com/wpcreditkt133.shtm

06-25-2001 04:20:54

New MessageRE:Memory Interleave Revisited (modified 0 times) Programmer
Profile
I updated the programs again.. I actually got the chance to test one on an IOpener over the weekend and found that when using the /pci option, it doesn't work. It does however work when using the /isa option, but due to a programming error when converting the original bi4v.bin to bi4v.com, (I left out the org 100h statement) it was pointing to the wrong place for the message, and I was using a jmp in the com2rom generator instead of a call so it wasn't setting the return on the stack properly. I will try the new programs on an IOpener tonight and let everyone know if they work in the morning.

I also updated http://bethie.net/~programmer/bi4v.com which you can run from dos to see exactly what should happen in the bios.

06-25-2001 08:27:14

New MessageRE:Memory Interleave Revisited (modified 0 times) ckbone
Profile
More memory Benchtests....Found two more ram chips. IBM/Hyundai 128meg, Crucial 64meg. Here are the results....very interesting.

IBM/Hyundai.....53/71 Identical to the Kingston chip. Best setting 2/fast.

Crucial (64meg).....62/86 Hard to figure...This score is almost as high as the PNY/Toshiba, in fact is the second fastest of all five chips I have....Never tested the stock ram chip. It actually feels faster in real world tests than the IBM or Kingston (128meg). This chip scored fastest with the 2/Turbo setting.

That does it....I'm buying a Crucial 128meg chip. Like I need another ram chip.....

06-25-2001 18:24:38

New MessageRE:Memory Interleave Revisited (modified 0 times) zyxw
Profile | Email
Hi,
A few comments please

1) THANK YOU for your effort!

2) Thank you!

3) Ok, now that you know I'm thankful... here's the critique

a) Everyone who changes bios's appears to like cbwin, well great, with trepidation I flash bios's and really prefer awdflash.exe... adding steps is both confusing, risky and annoying... may I suggest the bios be distributed in a complete format if possible?

b) I have had trouble with a V3 and V5 BIOS and the tweaked bios... hangs mostly with my USB HP CD ROM's... not sure if it's "usb filter" related, but when I flash back... no problem...

c) Currently I'm running an older only memory hacked bios and it seems kinda nice... the usb filter stuff doesn't appear to have helped...

Finally, well ok... deep down in my heart I know this ain't gonna happen, but has anyone figured out a way to boot off usb devices with an I-Opener??

Thanks!!

BTW... THANK YOU THANK YOU THANK YOU for your BIOS work... that I am having a problem with the USB CD ROM and the BIOS shouldn't be perceived as anything other than reporting my experience, and my desire for a simple (yup I'm a fool, I admit it) flashable directly BIOS is a total reflection of my own ignorance... and is my opinion not a fact

Thanks again!

06-25-2001 22:30:59

New MessageRE:Memory Interleave Revisited (modified 0 times) ckbone
Profile
More testing....Moving the Crucial 64mg chip between my three I-Openers achieved similar high benchmarks. Further testing revealed it's only flaw....not enough ram. On all my machines, the OS is adjusted to use only ram. Virtual memory is only used when all available ram is used up. This speeds up the system quite a bit....but is not reflected in any memory benchmark. So if you're only doing small tasks on the I-Opener, this 64mg Crucial chip performs well, better in fact than three of my 128mg chips. However, With Win 98 running, nothing else going, you only have about 8mg of free ram.....as soon as you start up a browser, that ram is gone and you immediately start using virtual memory (the hard drive) and your system slows down, proving that even a slow 128mg ram chip performs better in real world usage than a fast 64mg chip. Of course if your BIOS settings are off for your particular chip of any size, the whole System is f****d. One of my 128mg chips won't even boot if the ram timing is set to "turbo". It pays to benchmark the different BIOS settings.
06-27-2001 03:42:04

New MessageRE:Memory Interleave Revisited (modified 0 times) Programmer
Profile
IT IS .. ALIVE!!!!

OK maybe not.. but it DOES work. The bios at http://bethie.net/~programmer/v5pmi.bin has the (ISA) option rom http://bethie.net/~programmer/bi4v.bin installed with 'cbrom v5pmi.bin /isa bi4v.bin' and after flashing to a v5 IOpener it switched to text mode displayed a message, and proceded to actually let me see scandisk.

get 'em while they're hot...

06-27-2001 07:12:55

New MessageRE:Memory Interleave Revisited (modified 0 times) ckbone
Profile
Here's something else to think about while tweaking your memory settings.....In the Award BIOS screen, the timings are 8ns/10ns, normal, medium, fast, turbo. You would think there is a normal progression of settings here (8/10 the slowest, turbo the fastest). Amazingly this is not always the case with Award. On a number of Award BIOS settings, normal is faster than the fast setting, medium is sometimes the same as 8/10. I haven't been able to determine if this is the case with the I-Opener, mostly because of the close benchmarks with the different settings. If anyone knows anything for sure about this, let us all know.
06-27-2001 16:07:38

New MessageRE:Memory Interleave Revisited (modified 0 times) ratster
Profile
Programmer - I do believe you've done it! You truly do kick ass... :)

Now that I have a BIOS that may allow FreeBSD to be installed, I'm a happy man... I'll save that project for the weekend.

A virtual bag of Famous Amos Pecan Chocolate Chunk cookies and a gallon of 1% MooJuice (w/ some Kahlua for flavor :), all for you! Thanks!

06-27-2001 19:13:41

New MessageRE:Memory Interleave Revisited (modified 0 times) Georgie
Profile
Hi Everyone,

Just two notes.

Ckbone,

There are HUGE differences between the memory timings in the IO. I compared my ctcm16r - c't magazine's memory&cache benchmark program - scores with the 8ns and Turbo settings. Well, Turbo came out almost 20% (!!) faster. So, go ahead, and just test it, YMMV. My sytem is a K6-3+ running at 550MHz - 110 MHz FSB -, 128MB Memorytogo SODIMM.

zxyw,
A hint about your HP USB CD-ROM. Mine developed the same symptoms when I installed the DC/DC hack in my IO - which sped it up quite considerably. ;) Basically, whenever I boot at processor speeds in excess of 500MHz, the HP USB CD driver has extreme difficulties in loading. What you can see "from outside" is that Windows is constantly trying to load something - constant HDD LED activity -, but it just can't. The log file of course shows who the culprit is. So now I boot at 495, and jump it up to 550 (or 570) in Windows.

Hope this helps.

Geo

06-28-2001 19:17:44

New MessageRE:Memory Interleave Revisited (modified 0 times) ckbone
Profile
Georgie...I'm using Sandra 2001 for the memory testing. So far, I've run tests on five different chips, using all the settings on each chip, also transfering the chips between three different I-Opener versions, with different processors. A very long, time-consuming process. What I found was each chip performed differently with different timing settings. Often a "Turbo" setting was dramatically slower with a specific chip. Often when I progressed from 8ns to turbo, there an erratic progression of scores. Sometimes there were no changes at all. I'm not certain if the I-Opener BIOS has a standard progression.... some Award BIOS settings don't follow a logical order. And your own results may be dramatically different from mine. If you are satisified with your tests that Turbo is the best for you....use it. The point is....using benchmarks can help determine the best setting.
06-29-2001 04:08:21

New MessageRE:Memory Interleave Revisited (modified 0 times) ckbone
Profile
Here's the final benchmark (unless I buy a 256MB chip). This is a Crucial 128MB, 133mhz, Model # CT16M64S6W7E, direct from Crucial about $52. The benchmarks on this chip were the fastest of the six different chips I tried, only slightly, however.....ALU 64, FPU 89. This was on a V4, pentium 200mmx@210mhz, WinME, Memory interleave enabled, CL2/turbo setting.

What really surprised me doing this benchmarking, was that the Crucial 64MB chip scored almost as high as the 128MB chip...I would have thought more ram would yield higher scores. Real-world usage proved 128MB of any chip worked better than a fast 64MB chip. Ram is cheap right now.....a $20 PNY chip tested almost as fast as a Crucial chip. I can't tell much difference between them....

07-18-2001 15:48:18

Reply to Thread | Printer |
All times are PSTPowered by UltraBoard v1.62



Copyright © 2000, Netmake Inc. All Rights Reserved.
See Terms and Conditions for more information.




i-opener opener laptop notebook computer help drivers dll free windows dos repair fix linux mac macintosh 2000 95 98 nt pc configure hardware software sound video netscape explorer network networking lan wan software cmos fat bios printer card mouse modem ide scsi cd rom controllers scanner tape hard drive cgi scripts source code mp3