Looking for Award Flasher 5.00 or can I use UniFlash?

Discusses BIOS flashers and utilities from Award, AMI and Uniflash
cp
BIOS Guru
Posts: 1914
Joined: Mon Oct 21, 2002 9:07 pm
Location: Germany

it's not the number of memory banks of the sdram module you should be looking for but the number of banks of that the memory chips on the module are composed. as your module seems to be composed of 16Mbit chips the chipset should at least be able to do 2-way interleaving. just go ahead and give it a try. or post the chip id here so that someone can take a look at the datasheet.

2-2-5-2.0 1-way interleaved : 00010000b aka 10h
2-2-5-2.0 2-way interleaved : 00010001b aka 11h
2-2-5-2.0 4-way interleaved : 00010010b aka 12h
Alan_UK
BIOS Newbie
Posts: 16
Joined: Tue Dec 05, 2006 6:16 pm

cp wrote:i or post the chip id here so that someone can take a look at the datasheet.
Many thanks for that offer. According to Everest I have
1 x Hyundai 71V16635HCT8-H
2 x Infineon HYS64V16220GU-7.5

I tried a Google but hits were either in French or chip resellers.


I'm very much a newbie when we get down to this level but I've noticed the following from the Everest report:

1. Reporting 70 and 60 ns whereas the BIOS was reporting 8ns (now set to 10ns). I always thought 60/70 ns was SIMMs and 10ns was DIMMs

2. The Bank info is different for Banks 4 & 5 (2 x 64MB) compared to Banks 0 & 1 and 2 & 3 (both 1 x 128MB + 1 x empty) - is that because the Hyundai chip is different? If so should I get another Infineon? (the Hyundai was in the PC whereas I bought the Infineons)


Physically:
2 x SIMM slots - not used (manual says these are 1 Bank)
3 x DIMM slots (SRAM) with 128MB in each (manual says each is a Bank)

BIOS:
Bank 0/1 DRAM Timings SDRAM - default is 8ns and is set at 10ns
Bank 2/3 DRAM Timings SDRAM - ditto
Bank 4/5 DRAM Timings SDRAM - ditto

Everest:
Memory Controller Properties:
_____Error Detection Method None
_____Error Correction Single-bit
_____Supported Memory Interleave 4-Way
_____Current Memory Interleave 1-Way
_____Supported Memory Speeds 70ns, 60ns <<< compare to BIOS and chips
_____Supported Memory Types SPM, FPM, EDO, Parity, ECC, SIMM, DIMM, SDRAM
_____Supported Memory Voltages 5V, 3.3V
_____Maximum Memory Module Size 256 MB
_____Memory Slots 6

[ Memory Modules / BANK_0 ]
_____________Socket Designation BANK_0
_____________Type DIMM, SDRAM
_____________Speed 70 ns
_____________Installed Size 128 MB
_____________Enabled Size 128 MB

[ Memory Modules / BANK_1 ]
__________Memory Module Properties:
______________Socket Designation BANK_1
______________Type DIMM, SDRAM
______________Speed 70 ns
______________Installed Size Not Installed
______________Enabled Size Not Installed

[ Memory Modules / BANK_2 ]
______________Socket Designation BANK_2
______________Type DIMM, SDRAM
______________Speed 70 ns
______________Installed Size 128 MB
______________Enabled Size 128 MB

[ Memory Modules / BANK_3 ]
______________Socket Designation BANK_3
______________Type DIMM, SDRAM
______________Speed 70 ns
______________Installed Size Not Installed
______________Enabled Size Not Installed

[ Memory Modules / BANK_4 ]
______________Socket Designation BANK_4
______________Type DIMM, SDRAM
______________Speed 70 ns
______________Installed Size 64 MB
______________Enabled Size 64 MB

[ Memory Modules / BANK_5 ]
______________Socket Designation BANK_5
______________Type DIMM, SDRAM
______________Speed 70 ns
______________Installed Size 64 MB
______________Enabled Size 64 MB


Memory Bus Properties:
______________Bus Type SDR SDRAM
______________Bus Width 64-bit
______________Real Clock 100 MHz
______________Effective Clock 100 MHz
______________Bandwidth 800 MB/s

SPD Memory Modules:
DIMM1: Hyundai 71V16635HCT8-H 128 MB PC133 SDRAM (3.0-3-3-6 @ 133 MHz) (2.0-2-2-5 @ 100 MHz)
DIMM2: Infineon HYS64V16300GU-7.5 128 MB PC133 SDRAM (3.0-3-3-6 @ 133 MHz) (2.0-2-2-5 @ 100 MHz)
DIMM3: Infineon HYS64V16220GU-7.5 128 MB PC133 SDRAM (3.0-3-3-6 @ 133 MHz) (2.0-2-2-5 @ 100 MHz)


Currently running MemTest86+ just to check base setup is OK ..................
cp
BIOS Guru
Posts: 1914
Joined: Mon Oct 21, 2002 9:07 pm
Location: Germany

first: don't trust everest. i may be good for a rough guess on what is installed. instead use the datasheet and wcpredit to see what's really going on. according to this:
http://www.ortodoxism.ro/datasheets2/7/ ... toxw7y.pdf
the internal organisation of the memory modules on the hyundai module is 4bank. so you can use 4-way interleaving on that module. to test which memory banks belong to which northbridge offset just plug in one module and look which value the offsets hold. i suppose a D6h for an empty slot and something else for the slot with the module fitted. since you're running an 100MHz fsb you should set the memory to 'turbo' in the bios and check each memory module with memtest86+ and then with wcpredit for 4-way interleaving.
Alan_UK
BIOS Newbie
Posts: 16
Joined: Tue Dec 05, 2006 6:16 pm

cp wrote:according to this:
http://www.ortodoxism.ro/datasheets2/7/ ... toxw7y.pdf
the internal organisation of the memory modules on the hyundai module is 4bank. so you can use 4-way interleaving on that module.
Many thanks - that's a strange site to have a datasheet - seems to be about Greek Orthodox Religion. :?

I found a data sheet for the Infineon HYS64V16220GU-7.5 and I seem out of luck as it is only 1 bank:

http://www.datasheetarchive.com/datashe ... le=1714126
HYS 64V16300GU-7.5-C2 PC133-333-520 L-DIM-168-33 133 Mhz 16M × 64 1 bank SDRAM module

Currently I'm testing the memory - almost 3 passes and no errors.

Many thanks CP for your earlier post (the one you had to retype - that was kind of you).
cp wrote:you can set different timings for each bank (nice feature).
So, I can do it for 1 of my 3 sticks but it looks like I need to get different memory to do the whole lot - and here's me thinking that I was doing the right thing hunting down Infineon memory! :wink:

I guess if I were to do interleaving for 1 stick then it would be best to position that in Bank0/1 which I presume is the lowest memory that Windows would use first. Is that correct logic?
cp
BIOS Guru
Posts: 1914
Joined: Mon Oct 21, 2002 9:07 pm
Location: Germany

as i already wrote the datasheet for the infineon sdram stick mentions the organisation of the whole memory stick (1 bank). the internal organisation of the acutal chips is something completely different! but that's what you should look for. take the hyundai memory stick: it's a 1 bank stick (the whole thing) but the ram chips (each one of them) used to build it have an internal 4 bank organisation. that's how it's possible to use 4-way interleaving. again: interleaving only depends on the internal organisation of the ram chips not on the organisation of the memory stick.
the datasheets are only for the memory sticks, not for the ram chips. luckily hyundai added infos on the internal organisation of the used ram chips..infineon did not. so if you want to know if the infineon module support 4-way interleaving then you can: just switch it on with WCPREdit or post the markings on the ram chips (not the markings on the memory stick..they are mostly useless..)
which memory stick is bank0/1 and so on does NOT depend on windows in any way. it's determined how the traces on the mainboard are routed. consult the manual for more information. you should find something like DIMM0 = bank0/1 or just use one stick at a time and watch the data of the northbridge offsets holding the configuration data for the memory banks (64, 65 and 66).
Denniss
BIOS Guru
Posts: 3153
Joined: Thu Mar 21, 2002 8:16 pm
Location: Near Hannover (CEBIT) Germany
Contact:

This Infineon datasheet is somewhat hard to understand but it looks like they have either 2-way or 4-way interleave.
Alan_UK
BIOS Newbie
Posts: 16
Joined: Tue Dec 05, 2006 6:16 pm

cp wrote:again: interleaving only depends on the internal organisation of the ram chips not on the organisation of the memory stick.
Got it - penny has dropped (as we say in the UK) - or light bulb has lit :idea:

Thanks for your patience, cp and Denniss.

4+ hours of memory test did 3 cycles and no errors. Time to try a little experimenting................
KachiWachi
The New Guy
Posts: 1451
Joined: Fri Mar 29, 2002 10:32 pm
Location: Pennsylvania, USA

@cp - correct...it is the "banking" of the chips themselves. The OP would have to provide that information. I guess I read the datasheet one way, and made assumptions based on that...as compared to other datasheets I have read.

Everest will only show the interleave for the last bank it reads IIRC. I had previously PM'd Fiery about this over at Lavalys, and had a lengthy PM about it. Suffice to say that they will not post the interleave for each populated bank. :(

WPCREDIT will show what the system is setting up for each RAM bank...though you will have to decode it manually using the chipset datasheet.
Alan_UK
BIOS Newbie
Posts: 16
Joined: Tue Dec 05, 2006 6:16 pm

Alan_UK wrote:4+ hours of memory test did 3 cycles and no errors. Time to try a little experimenting................
Slight delay in plan. Have just bought another 128MB on eBay. Seller said it was a Hyundai HY57V28820AT-H which I think is OK for interleaving. Data sheet at http://www.datasheets.org.uk/datasheet. ... le=1700950 says
"single control command (Burst length of 1,2,4,8 or full page), and the burst count sequence(sequential or interleave). ..... 4Banks x 4Mbits x 8"

Another eBay Hyundai chip HY57V28820HCT-H is due to end tomorrow.
Data sheet says the same as the HY57V28820AT-H.
http://www.datasheets.org.uk/datasheet. ... le=1700971


One thing that concerns me. If the bios setting set interleave and faster speed and other changes, do the PCI boards (particularly the ATA1-100) cope OK with the different memory settings? As I understand it, DBA means direct PCI to Memory. Perhaps as everything is going via the NorthBridge, it ensures it's all OK.
cp
BIOS Guru
Posts: 1914
Joined: Mon Oct 21, 2002 9:07 pm
Location: Germany

yeah, now you've got it right :) those are the datasheets for the ram chips used to build the memory stick. the datasheets clearly indicate 4bank, so memory sticks build of those are 4-way interleavable.

regarding your concerns about the pci devices: no problem at all. first of all different speed settings for each memory bank is a feature of the (VIA) northbridge. there are other northbridges from other manufacturers that only allow a single setting for all memory banks. as long as you use identical modules on those northbridges nothing will happen, but if you use different modules (in terms of access speed) problems (may) arise. there are no problem like this even with different modules on this northbridge, because it (should) set the optimal speed for each bank. about the pci devices: they don't even know to which module they write. they only know the address..the rest is done by the northbridge. even in DMA mode. DMA only says that there's no CPU interaction needed to transfer data from one device to another (memory is a device, too). to start DMA a source and a destination device, a start address and the number of blocks to transfer is given to the northbridge by the CPU. the rest is done without the cpu...so it has some time to do more important things ;)
or short: the northbridge handles the timing for every memory write. there is no way around the northbridge if you want something from the ram.
Alan_UK
BIOS Newbie
Posts: 16
Joined: Tue Dec 05, 2006 6:16 pm

cp wrote:or short: the northbridge handles the timing for every memory write. there is no way around the northbridge if you want something from the ram.
Hi, cp, many thanks for that clear explanation. It seems that the Northbridge is a key but highly neglected component of a PC - all that focus on the processor and it's speed. Let's face it, historically computers mainframe and were all about number crunching but PCs are often more about interaction with the user (video) and information (data on hard disk and increasingly on the internet), and it all goes through the northbridge!
cp
BIOS Guru
Posts: 1914
Joined: Mon Oct 21, 2002 9:07 pm
Location: Germany

by accident i stumbled across the SPD JEDEC standard (again) and took a closer look at some of the contents..and guess what? there's a byte that indicates the number ob banks of the ram chips used on the memory module. just grab cpu-z and use the 'registers dump' button at the tab 'about'. you'll find a dump of all memory modules' spd data in that file, too. now check byte 11h (that's byte 17d). the number in there indicates the number of banks of the ram chips e.g. 01 for 1 bank, 02 for 2 banks, 04 for 4 banks, etc. now you know which memory interleaving option is possible for each module. please note that some (mostly cheap) memory modules _may_ contain bogus spd data.

references:
JEDEC SPD Specification
CPU-Z
Alan_UK
BIOS Newbie
Posts: 16
Joined: Tue Dec 05, 2006 6:16 pm

cp wrote:just grab cpu-z and use the 'registers dump' button at the tab 'about'. you'll find a dump of all memory modules' spd data in that file, too. now check byte 11h (that's byte 17d). the number in there indicates the number of banks of the ram chips
Many thanks cp.

I've at last found some more memory chips on ebay that allow interleaving at a reasonable price (there were too many snipers paying over the odds!). I've installed Prime85, memtest86 and WCPREdit so ready to go when they arrive, maybe after Christmas...


A a happy holiday season to you, KachiWachi and Denniss
KachiWachi
The New Guy
Posts: 1451
Joined: Fri Mar 29, 2002 10:32 pm
Location: Pennsylvania, USA

Great catch cp! :wink:

I think Everest shows this as well (SPD page).
Alan_UK
BIOS Newbie
Posts: 16
Joined: Tue Dec 05, 2006 6:16 pm

Hi cp, some progress, some disappointments.

Using BIOS settings

1. Changed memory speed to Turbo but MemTest86 reported errors

2. Changed memory speed to Fast but MemTest86 reported errors

3. Changed memory speed to 8ns and MemTest86 reported OK (8 hours)

4. Changed memory Cycle Length to 2 (from 3) and MemTest86 reported OK (8 hours)

Improved timings, especially on Win98 loading: - all min:secs

Load Windows 98: 1:45 (from 2:20)
Load FireFox: 0:20 (from 0:25)
Load Acrobat: 0:30 (from 0:23)

But hardly any change with Everest Reports:

Memory Read: 216 MB/s - same
Memory Write: 90 MB/s - same
Memory Latency: 323.5 ns - was 325.6


Using WPCREDIT and then WPCRS120 to permanently set changes:

5. Set all 3 memory banks to interleave (offsets 64/65/66) from D4 to D6 (i.e. set right most 2 bits from 00 to 10 (4-way). Ran Prime95 for 5.5 hours with no errors or warnings.

No change on the sample Load times or Everest Memory timings.

cp wrote:just grab cpu-z and use the 'registers dump' button at the tab 'about'. you'll find a dump of all memory modules' spd data in that file, too. now check byte 11h (that's byte 17d). the number in there indicates the number of banks of the ram chips
I tried this but the program reports no SPD data available (I'm using the following as reported by Everest)

DIMM1: Hyundai 71V16635HCT8-H 128 MB PC133 SDRAM (3.0-3-3-6 @ 133 MHz) (2.0-2-2-5 @ 100 MHz) (1 rank, 4 banks)
DIMM2: C Memory DP133-064163E 128 MB PC133 SDRAM (3.0-3-3-6 @ 133 MHz) (2.0-2-2-5 @ 100 MHz) (1 rank, 4 banks)
DIMM3: Micron Tech. 16LSDT1664AG-133C7 128 MB PC133 SDRAM (3.0-3-3-6 @ 133 MHz) (2.0-2-2-5 @ 100 MHz) (2 ranks, 4 banks)



Maybe I'm getting to the end of the road or maybe something has not been set correctly.

I would appreciate any advice as whether to change the following memory control bits in offset 64/65/66 as reported by WPCREDIT:

1. Active to CMD to 2T - currently 3T
2. DDR Write enable - currently disabled
3. Precharge to Active to 2T - currently 3T
4. Active to Precharge to 5T - currently 6T

or as reported by Everest:

Memory Timings:
CAS Latency (CL) 2T
RAS To CAS Delay (tRCD) 3T
RAS Precharge (tRP) 3T
RAS Active Time (tRAS) 6T

and Everest says the memory supports:

Auto-Precharge Supported
Precharge All Supported
Write1/Read Burst Supported

Many thanks for your continuing help and support.
Post Reply