HELP!!!!!! AMD DX4-100 Speed Problem!!!!
Steve Jackson
wrote:
>HELP!!!!!!
>
>I have recently bought a new MB and AMD DX4-100 processor and am
>experiencing low CPU benchmark results in Norton 8.0 (130.6). Also if
>I try an AMD DX2-66 in the same MB I get an identical benchmark result
>(but get a greater result in an older MB!) I am 99.9% sure that all
>the jumpers on the MB are correct but there are so many CMOS settings
>that I don't really know where to start!
>
> Here are some of the settings that I have set :-
>
>System Boot up cpu speed = high
>external cache = enabled
>internal cache = enabled
>internal cache WB/WT = Write-thru
>cache read hit wait state = 3-1-1-1
try = 0ws
>cache write hit wait state = 1ws
and =0ws
>dram wait state select =1ws
>Dram page mode= fast
>AT clock select= PCLK/4
>local ready delay setting= delay 1T
>CPU ADS# delay 1T or not= no delay
>Alt bit in tag sram= 7+1 Bits
>signal LDEV# sample time= in T3
>ISA BUS refresh disable bit= fast
>Divider for refresh= 1/16
>LOWA20# select= chipset
>RC reset select= chipset
>IO recovery time select= 5 BCLK
>
>Do I need a different type of MB?
>Has WB/WT cache anything to do with it?
>
>Can anybody help!?!
>
>Thanks
>
I had exactly the same thing this weekend ( Tell me about it !!)
It really pissed me off, but i think i cracked it !
Hope this helps!
Jacko..
Jim Stiel
difference) and set the cache to 2-1-1-1
Mark Dixon
I have recently bought a new MB and AMD DX4-100 processor and am
experiencing low CPU benchmark results in Norton 8.0 (130.6). Also if
I try an AMD DX2-66 in the same MB I get an identical benchmark result
(but get a greater result in an older MB!) I am 99.9% sure that all
the jumpers on the MB are correct but there are so many CMOS settings
that I don't really know where to start!
Here are some of the settings that I have set :-
System Boot up cpu speed = high
external cache = enabled
internal cache = enabled
internal cache WB/WT = Write-thru
cache read hit wait state = 3-1-1-1
cache write hit wait state = 1ws
Uwe Bochert
> I have recently bought a new MB and AMD DX4-100 processor and am
> experiencing low CPU benchmark results in Norton 8.0 (130.6). Also if
> I try an AMD DX2-66 in the same MB I get an identical benchmark result
> (but get a greater result in an older MB!) I am 99.9% sure that all
> the jumpers on the MB are correct but there are so many CMOS settings
> that I don't really know where to start!
[snip]
A friend have the same problem, he find out that he has bought a
486dx2/66 with a 486dx4/100 label. This is like the placebo cache
in some motherboards, fake what a f$%'!!!! For me it looks like
the market is flooded with faked CPUs and CacheRAMs.....
Maybe you read about this in some comp-mags. Ask the CPU: 'who are
you?' with some progs and if the CPU tell you 'I m a 486dx2/66' you
should immediatly contact your dealer. :-[
--
(o)(o)
< .. >
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Uwe Borchert *** eMail ub...@rz.uni-karlsruhe.de
Steve Jackson
wrote:
>I would suggest setting the CMOS to Write Back (yes, it can make a big
>difference) and set the cache to 2-1-1-1
>
>
>
>
I use System Info Version 9.0
and i only get 104 on the cpu benchmark and Disk speed test is only
600k/bs
I have a Samsung 1.2 Gig with Eide selected in the BIOS, so why so
slow ?
I have a AMI BIOS 1994 and i cant find anything to do with setting
the cache to 2-1-1-1
My setting are :
AT BUS clock CLK2/4 ( What the hell does this mean?)
DRAM READ TIMING Normal ( I tried fast and it locks up)
DRAM WRITE TIMING Normal ( " )
SRAM READ TIMING 0ws ( When i put these to 1 the computer slows
to 66 Mhz in S.I. Benchmark)
SRAM WRITE TIMING 0ws
CYCLE CHECK POINT Faster ?
Ext Cache WB ( I have 256K Cache)
Int Cache WB
Can anyone help me with the optimal settings for a AMDdx4 100 in a ALI
PCI motherboard ?
Or should i contact AMD, AMI or ALI ( What does ALI stand for ?)
Anyone know any addresses
Thanks to everyone for the help !!
Jacko.
Steven J. Devine
>Mark Dixon (m...@madsoft.demon.co.uk) wrote:
>> HELP!!!!!!
>> I have recently bought a new MB and AMD DX4-100 processor and am
>> experiencing low CPU benchmark results in Norton 8.0 (130.6). Also if
>> I try an AMD DX2-66 in the same MB I get an identical benchmark result
>> (but get a greater result in an older MB!) I am 99.9% sure that all
>> the jumpers on the MB are correct but there are so many CMOS settings
>> that I don't really know where to start!
>A friend have the same problem, he find out that he has bought a
>486dx2/66 with a 486dx4/100 label.
Another potential problem (a little more benign than the above) is that there
MUST BE a jumper on the motherboard to tell the CPU that it is running in 3x
mode, as opposed to 2x mode. If the jumper is set to 2x, or there is no
2x/3x jumper on the motherboard, the CPU will run exactly as a 66 (2x33), and
score the same benchmark as a 66. If it were a 66 in disguise, it would not
run in 3x mode, though I do not know what the effect would be if there was a
66 in a socket jumpered for 3x.
Steve
---------------------------- Steven J. Devine -----------------------------
--------------- personal account: dev...@tiac.net -------------------------
------------------- work account: dev...@mapo1.him.unisys.com -------------
ma...@interlog.com
>HELP!!!!!!
>
>I have recently bought a new MB and AMD DX4-100 processor and am
>experiencing low CPU benchmark results in Norton 8.0 (130.6). Also if
>I try an AMD DX2-66 in the same MB I get an identical benchmark result
>(but get a greater result in an older MB!) I am 99.9% sure that all
>the jumpers on the MB are correct but there are so many CMOS settings
>that I don't really know where to start!
>
***edited for bandwidth***
>Can anybody help!?!
>
>Thanks
Just a shot in the dark, but have you checked the TURBO setting on
the motherboard? My experience is some are backwards and do not
display correctly on the LED. Try reversing the leads or just push the
button in or out depending. Of course this is if there is a turbo
switch.
I wonder about the WB/Writethru as well. Mybe some one will post.
Michael Carter [ma...@interlog.com]
Michael Kelcy
says...
>
>HELP!!!!!!
>
>I have recently bought a new MB and AMD DX4-100 processor and am
>experiencing low CPU benchmark results in Norton 8.0 (130.6). Also if
>I try an AMD DX2-66 in the same MB I get an identical benchmark result
>(but get a greater result in an older MB!) I am 99.9% sure that all
>the jumpers on the MB are correct but there are so many CMOS settings
>that I don't really know where to start!
>
>
>System Boot up cpu speed = high
>external cache = enabled
>internal cache = enabled
>internal cache WB/WT = Write-thru
>cache read hit wait state = 3-1-1-1
>cache write hit wait state = 1ws
>dram wait state select =1ws
>Dram page mode= fast
>AT clock select= PCLK/4
>local ready delay setting= delay 1T
>CPU ADS# delay 1T or not= no delay
>Alt bit in tag sram= 7+1 Bits
>signal LDEV# sample time= in T3
>ISA BUS refresh disable bit= fast
>Divider for refresh= 1/16
>LOWA20# select= chipset
>RC reset select= chipset
>IO recovery time select= 5 BCLK
>
>Do I need a different type of MB?
>
>
>Thanks
>
Mark: I am by no means a techie, but the rule I follow is fewer wait states
mean faster (albeit at some point), less stable PC. Try changing settings in
the BIOS (one at a time is best if you have the time to play with it) to fewer
wait states and then run your favorite benchmark. Eventually it should be much
faster.
William Mitchell
>Hello Mark, hello rest of the world,
>
>Mark Dixon (m...@madsoft.demon.co.uk) wrote:
>> HELP!!!!!!
>> I have recently bought a new MB and AMD DX4-100 processor and am
>> experiencing low CPU benchmark results in Norton 8.0 (130.6). Also if
>> I try an AMD DX2-66 in the same MB I get an identical benchmark result
>> (but get a greater result in an older MB!) I am 99.9% sure that all
>> the jumpers on the MB are correct but there are so many CMOS settings
>> that I don't really know where to start!
>[snip]
>A friend have the same problem, he find out that he has bought a
>486dx2/66 with a 486dx4/100 label. This is like the placebo cache
>in some motherboards, fake what a f$%'!!!! For me it looks like
>the market is flooded with faked CPUs and CacheRAMs.....
>Maybe you read about this in some comp-mags. Ask the CPU: 'who are
>you?' with some progs and if the CPU tell you 'I m a 486dx2/66' you
>should immediatly contact your dealer. :-[
>--
> (o)(o)
> < .. >
>~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>Uwe Borchert *** eMail ub...@rz.uni-karlsruhe.de
Just a shot in the dark. Check the setting for DRAM speed in the
CMOS. My Award 4.50G bios has this setting. If set to fast, the
system performs like a DX2-66. This setting must be set to fastest.
William Mitchell
email: wmit...@escape.ca
Uwe Bochert
you (wmit...@escape.ca) wrote:
> >Mark Dixon (m...@madsoft.demon.co.uk) wrote:
> >> HELP!!!!!!
> >> I have recently bought a new MB and AMD DX4-100 processor and am
> >> experiencing low CPU benchmark results in Norton 8.0 (130.6). Also if
> >> I try an AMD DX2-66 in the same MB I get an identical benchmark result
> >> (but get a greater result in an older MB!) I am 99.9% sure that all
> >> the jumpers on the MB are correct but there are so many CMOS settings
> >> that I don't really know where to start!
> >[snip]
> >A friend have the same problem, he find out that he has bought a
> >486dx2/66 with a 486dx4/100 label. This is like the placebo cache
> >in some motherboards, fake what a f$%'!!!! For me it looks like
> >the market is flooded with faked CPUs and CacheRAMs.....
> >Maybe you read about this in some comp-mags. Ask the CPU: 'who are
> >you?' with some progs and if the CPU tell you 'I m a 486dx2/66' you
> >should immediatly contact your dealer. :-[
> >Uwe Borchert *** eMail ub...@rz.uni-karlsruhe.de
> Just a shot in the dark. Check the setting for DRAM speed in the
> CMOS. My Award 4.50G bios has this setting. If set to fast, the
> system performs like a DX2-66. This setting must be set to fastest.
Sorry, but the jumpers were ok and he tested a real dx4/100 with the
same jumpersetting und get correct benchmark results. It looks like
this CPUs are realy dx2/66 with faked labels as you can read sometimes
in the newspapers and comp mags. I will give your information to my
friend, but i got no real hope, its just to be 200% sure!!!
Thanx a lot, ....
Jeff Jonas
>In article <8167809...@madsoft.demon.co.uk>, m...@madsoft.demon.co.uk
>says...
>>HELP!!!!!!
>>I have recently bought a new MB and AMD DX4-100 processor and am
>>experiencing low CPU benchmark results in Norton 8.0 (130.6).
>Mark: I am by no means a techie, but the rule I follow is fewer wait states
>mean faster (albeit at some point), less stable PC. Try changing settings in
^^^^^^^^^^^^^^
>the BIOS (one at a time is best if you have the time to play with it)
>to fewer wait states and then run your favorite benchmark.
>Eventually it should be much faster.
THAT IS EXACTLY MY POINT - WHY is it trial and error when
you ought to be able to match spec sheets and have the correct
settings THE FIRST TIME!
Why is it so hard to get the specs for this tuning, considering
that you BOUGHT AND OWN THE HARDWARE, unlike software "licences"?
--
Jeffrey Jonas
je...@panix.com
WILLIAM C YU
: Sorry, but the jumpers were ok and he tested a real dx4/100 with the
: same jumpersetting und get correct benchmark results. It looks like
: this CPUs are realy dx2/66 with faked labels as you can read sometimes
: in the newspapers and comp mags. I will give your information to my
: friend, but i got no real hope, its just to be 200% sure!!!
I would say the motherboard is problably the culprit. The problem is that
the AMD and Cyrix CPUs can operate in either 2X or 3X mode. That means if
your motherboard does not have the appropriate clock-selector options
and the clock-selector option is set to 2X by default, then the DX4/100
will run simply as a DX2/66.
You have a few possible options. You can try using the DX2/100 mode. Put
the clock selecter to 2X but up the bus speed to 50MHz. (Note, this mode
performs better than the default 33MHz x 3.) If this doesn't work, then
you have no choice but to either upgrade the BIOS or buy a new motherboard.
Hmmm, I suppose you could try snapping off the pin that controls the
clock selection on the AMD CPU but I doubt this solution has much appeal.
Intuitive Microsystems
>HELP!!!!!!
>I have recently bought a new MB and AMD DX4-100 processor and am
>experiencing low CPU benchmark results in Norton 8.0 (130.6). Also if
>I try an AMD DX2-66 in the same MB I get an identical benchmark result
>(but get a greater result in an older MB!) I am 99.9% sure that all
>the jumpers on the MB are correct but there are so many CMOS settings
>that I don't really know where to start!
Check and recheck the jumper settings on the motherboard related
to the CPU type. The AMD DX4/100 needs at least 1 jumper change to
make it run in 3 x clock mode. Otherwise it runs in 2 x clock mode.
What's the difference?
2 x 33MHz clock = 66MHz
3 x 33MHz clock = 99MHz (this number is rounded up to 100MHz)
Watch your system as it boots, it is probably reporting the AMD
DX4/100 as a DX2/66.
L. L. Kilgore
>m...@madsoft.demon.co.uk (Mark Dixon) wrote:
>> I am 99.9% sure that all the jumpers on the MB are correct
>Maybe you should not be so sure. I had behavior just like you describe
>when I set mine up a couple of weeks ago. It turned out that a single
>jumper was missing. Even though the startup screen said it was running
>at 100, it was really only running at the equivalent of 66. Adding
>that one jumper took it from a Norton SI of about 135 to just under
>200.
>BTW - I didn't see any really significant changes from fiddling with
>CMOS settings - except that some of them kept it from being able to
>boot up :-)
The new 486 VLB MB with enhanced AMD DX4/120 that I installed
yesterday had 6 missing or misplaced jumpers. The CMOS settings
weren't exactly optimal either, but I couldn't make much of an
improvement there until I'd corrected the jumper problem. The
settings I'm using tonight wouldn't allow the machine to boot last
night.
Laura
Ed Hochman
jas...@cse.unsw.edu.au (Jason Joseph Lau) wrote:
>Steven J. Devine (dev...@tiac.net) wrote:
>
>I've heard reports that some people have successfully installed DX4's in
>motherboards that do not support 3x, and had it working at full speed.
>
>It may not report it as a DX4, but they say it ran at full DX4-100 speed...
>
>Finding such a jumper as you describe would be difficult for people
>people like me who have motherboards which accept DX-33's
>and DX2-66's for example, on the same jumper setting. Somehow the
>BIOS is able to detect that my chip is a DX2-66 and not a DX-33!
>
>Going by this, I may be able to fit a DX4 chip into my system, and
>have it running at 100Mhz internally, without the BIOS actually having
>to acknowledge that it is a DX4 chip. (My BIOS was manufactured
>before the onslaught of DX4's)
>
>I really want to upgrade to a DX4, so can anybody confirm the above claim
>with some solid facts? I have a MR.BIOS V1.53 and an OPTI 82C611 chipset.
>
I don't think there is any jumper to identify the cpu as a 2x or 3x. The
whole idea of the clock doubling/trippling is that it's done on the chip; not
the mother board. As far as the mb is concerned, dx2/66's and dx4/100's are
both dx33's; and they do communicate with the rest of the board at 33mh.
However, the key concern I'd have is that your mb may not support the voltage
requirements of the dx4. Most dx's and dx2's are, I think, 5 volt devices;
and most dx4's are, I think, 3.3v. MB's I've seen that support dx4's have on
board voltage regulators and a jumper that let's you select the voltage
supplied to the cpu. I suspect you could ruin the chip by feeding it too high
a voltage.
Ed
Ed Hochman MBH Systems, Inc. e...@mbhsys.com
john mitchell
The Dx-4-100 repaces any 33 chip eg 33 or 66. It is clock trippled so runs
automatically at 100. It was just a plug and play for me.
john mitchell
john mitchell
Dx4-100 overdrives are 5 volt chips -have a voltage reducer built in.
john mitchell
Mike Russell
Your board must know about the AMD DX4 or you will just get a 2X
multiplier. You cannot just set the board up for an Intel DX4/100
with a 3X multiplier - there must be a specific jumper for the AMD
DX4.
BTW I doubt very much that your chip was mislabelled or is counterfeit.
Scott Sterling
: The AMD DX4/100 uses a different pin to select the 3X multiplier.
: Your board must know about the AMD DX4 or you will just get a 2X
: multiplier. You cannot just set the board up for an Intel DX4/100
: with a 3X multiplier - there must be a specific jumper for the AMD
: DX4.
Actually, the chip on most motherboards defaults to 3X. If it doesn't
then it is safe to move the bus speed up to 50. But before doing that,
the bios must be tweaked (to make sure it is only doubled). One wait
state on the cache will do amazingly bad things to the processing speed.
--Scott
Tom
: Uwe Bochert (ub...@rzstud1.rz.uni-karlsruhe.de) wrote:
A better choice would be to buy the Evergreen voltage converter (part
number 273). It will work with 3v motherboards and 5v motherboards, but
also includes a jumper to select 2x or 3x. It costs $39-$49, but it is
also in very short supply (nice business to be in!).
--
*************************************************************************
******************* Internet: tc...@vnet.net *******************
******************* Fidonet: 1:3634/12.500 @fidonet *******************
*************************************************************************
KEhrhardt
internally. The CLKMUL pin controls the internal clocking of the
processor. This pin is B13 on the processor pin out. Pull this pin to 5V
and the processor clocks at 3x, pull this pin to GND and the processor
clocks at 2x. Does this mean you can run a 486DX2-66 at 100MHz. You bet.
Reduce the external crystal frequency to 33MHz and jump B13 to 5V and it
runs at 100MHz. Now you have to handle the thermal problem. I use a CPU
cooler plus a second case fan blowing on the CPU cooler. Haven't had any
problems with this setup. I've tried 2x 50MHz with a 486DX2-80, however
in high ambients the system would hang.
Dave Mason
>jo...@cruzio.com (John Bandhauer) wrote:
>Laura
I'm thinking of buting a DX4-120 and motherboard this christmas, too.
Lemme know what sort of results you're getting.
thanx,
Dave
Michael Chan
>The AMD DX4/100 uses a different pin to select the 3X multiplier.
It was the story of the old version AMD DX4/100. The new Enhanced version
already pin-compatible Intel DX4/100 in this issue.
Regards,
Michael
Yak Aik Seng
My friends and I have exactly the same problems with the "normal" AMD
DX4-100 chip. Our chips function at 100Mhz on the (tested on the SYSINFO
and QAPLUS), but the CPU speed (NU Sysinfo CPU index = 130) is equal or
worse than a DX2-66 chip. We have swap boards, swap chips, changed jumpers,
switched BIOS settings etc you name it.
I have come to the following findings to report:-
(My motherboard has an UMC chip set with supposedly 256Kb external cache)
a) changing CPU speed to 50Mhz x 2 on my motherboards won't work
b) changing to Enhanced AMD DX4-100 seems to work to its reported
speed on the SYSINFO program (cpu index = 216).
c) changing the motherboard to an OPTI chipset with the "normal"
DX4-100 chip improves the speed to its reported speed (sysinfo
cpu index = 197.1)
d) setting the clock jumper my motherboard to 2X brings the
clock speed down to 66Mhz and sysinfo cpu index = 130, this
is equivalent to DX2-66 performance.
e) setting the BIOS with external cache disable slows down the
sysinfo index to about 70. Other BIOS settings will not improve
the system significantly.
This is my conclusions:-
SOME of the external cache memories used on my motherboard are
pacebos. It may be reported as 256Kb external cache memory when
detected by the BIOS, BUT I believe the board manufacturers
CHEATED(!!) the WORLD(!!) by putting in, perhaps, only 128Kb.
Please do not get me wrong, I am not saying that the OPTI
chipset is good and the UMC chipset is bad. I am referring
to some of those unscrupulous board manufacturers.
Greg Miller
>This is my conclusions:-
> SOME of the external cache memories used on my motherboard are
> pacebos. It may be reported as 256Kb external cache memory when
> detected by the BIOS, BUT I believe the board manufacturers
> CHEATED(!!) the WORLD(!!) by putting in, perhaps, only 128Kb.
I have several systems which perform well below what they should.
All of these systems have 256k of cache. However, today when I ran
WinProbe on one of the systems, it reported 0% Cache hits. I then ran
this program on all my systems which were performing below average,
and they too reported 0% cache hits.
My systems which perform normally produce between 60% and 80%
cache hits.
I'll look into your suggestion that there really isn't 256k of
cache there by swapping the cache in a system which really works with
one of the slow systems.
Let me know if you (or anyone else) comes up with anything else
along these matters. (BTW: the boards are of several different
types, as well as the CPUs).
Hey Santa, how much for your list of naughty little girls?
greg....@shivasys.com
http://www.ius.indiana.edu/~gmiller/
http://www.shivasys.com/greg/
Michael Finkel
: Hi all,
: My friends and I have exactly the same problems with the "normal" AMD
: DX4-100 chip. Our chips function at 100Mhz on the (tested on the SYSINFO
: and QAPLUS), but the CPU speed (NU Sysinfo CPU index = 130) is equal or
: worse than a DX2-66 chip. We have swap boards, swap chips, changed jumpers,
: switched BIOS settings etc you name it.
Well, I am having the same problems with an Intel DX4-100 on a Soyo
486PCI board (based on the AMI bios and the ALI chipset). I am getting
sysinfo CPU scores of 132.3 consistently -- which is worse than my old VL
DX2-66 setup which got ~140. If I change the SRAM read cycle to 2-1-1-1
from the 3-1-1-1 that is auto-detected, performance jumps to 198.5 which
is what I expect and want to see, but the system crashes after a shirt
while in windows or DOS...
I am contemplating swapping the motherboard for an ASUS sp3g, but I don't
know if that will make things any better.
This sucks..
any suggestions?
Tim Pace
>
>mpey...@leonis.nus.sg (Yak Aik Seng) wrote:
>
>>This is my conclusions:-
>
>> SOME of the external cache memories used on my motherboard are
>> pacebos. It may be reported as 256Kb external cache memory when
>> detected by the BIOS,
Just a note of caution. Typically the BIOS checks the cache jumper
settings, not the actual L2 cache installed. Try resetting the jumpers
to see for yourself. And I too purchased a board with fake cache. I
had to [ark myself in the retailer's store, listening to all his
bologna, until I got the fake chips swapped out. Worked fine after that.
Darren Rhodes
>This is my conclusions:-
> SOME of the external cache memories used on my motherboard are
> pacebos. It may be reported as 256Kb external cache memory when
> detected by the BIOS, BUT I believe the board manufacturers
> CHEATED(!!) the WORLD(!!) by putting in, perhaps, only 128Kb.
> Please do not get me wrong, I am not saying that the OPTI
> chipset is good and the UMC chipset is bad. I am referring
> to some of those unscrupulous board manufacturers.
Yep, I had the same with a UMC chipset m/b and now with a 486 Deep
Green m/b. The SI score is 129.5 with the AMD chip and about 132.0
with the Intel DX4-100 chip (The Intel has 16K level 1 cache). All
these scores are representative of machines with no level 2 cache.
I phoned the supplier about my second board and he said, "No it
doesn't have any cache ram as it uses write back technology". I didn't
bother arguing further.
P.S. At least I'm not the only to have been conned!!!