AMD 5x86-P75 (X5-133) with PCI/I-SP3G
hfreitag
>It is possible to use the AMD 5x86-P75 133 with the
>PCI/I-SP3G Mainboard ?
>
Have a look to the 3/96 c't magazin.
There's a lot of information about this. Not only the ASUS PCI/I SP3G
is covered.
Ronald F. Guilmette
Could someone PLEASE post the relevant information from that article (in
English) to this newsgroup!
It would help a lot, as I am about to try this combination of motherboard
and CPU.
--
-- Ron Guilmette, Roseville, CA -------- Infinite Monkeys & Co. ------------
---- E-mail: r...@monkeys.com ----------- Purveyors of Compiler Test Suites -
------ Copyright (c) 1996 by Ronald F. Guilmette; All rights reserved. -----
NO PEEKING!!
=>In article <4g76db$j...@grimsel.zurich.ibm.com>, <hfre...@vnet.ibm.com> wrote:
=>>In <4fq6de$18...@rs18.hrz.th-darmstadt.de>, Klaus Kaltwasser <kal...@hrz.th-darmstadt.de> writes:
=>>>It is possible to use the AMD 5x86-P75 133 with the
=>>>PCI/I-SP3G Mainboard ?
=>>>
=>>Have a look to the 3/96 c't magazin.
=>>There's a lot of information about this. Not only the ASUS PCI/I SP3G
=>>is covered.
=>>
=>Could someone PLEASE post the relevant information from that article (in
=>English) to this newsgroup!
=>It would help a lot, as I am about to try this combination of motherboard
=>and CPU.
=>--
=>-- Ron Guilmette, Roseville, CA -------- Infinite Monkeys & Co. ------------
=>---- E-mail: r...@monkeys.com ----------- Purveyors of Compiler Test Suites -
=>------ Copyright (c) 1996 by Ronald F. Guilmette; All rights reserved. -----
--It's already been done! And my thanks to the generosity of Ulrich Teichert
for providing that valuable piece of information. For those who have missed
it, here it is again....
Hi!
First, I know I should not crosspost, sorry. But I was asked to do
a translation of this test in alt.comp.periphs.mainboard.asus and this
group isn't available on many sites, but comp.sys.ibm.pc.hardware.chips
is. So, I'm guilty. Flame me if you like.
Disclaimer: this is not an ad, nor I got payed for the translation.
All data is from c't 3/96, pages 192-203.
If you spent money or destroy you hardware based on data
of this text, wether I nor c't are responsible, etc., bla bla
Comparison of the various CPUs in the test (WT=write through, WB=write back):
type cost fre.(MHz) L1-ca Vol Pinout Powerman L1-c-rw Int-mul
(DM) (ext/int) (KB) (cyc.) (cyc.)
iDX2-66 198 33/66 8/WT 5 iDX/DX2 none/iS 1 13
iODPR486DX33 219 33/66 8/WT 5 iDX/DX2 none/iS 1 13
iDX4-100 147 33/100 16/WT 3.45 iDX4 iS 1 5
iDX4-100WB n.a. 33/100 16/WB 3.3 iDX4-WB iS 1 5
iDX4ODPR 259 33/100 16/WT 5 iDX/DX2 iS 1 5
AMD486DX4-100SV8B 137 33/100 8/WB 3.45 iDX4-WB iS 1 13
AMD486DX4-120SV8B 198 40/120 8/WB 3.45 iDX4-WB iS 1 13
Cy Cx486DX4-100GP n.a. 33/100 8/WB 3.45 CyrixDX4 Cyrix 2 3
Cy Cx486DX4-100GP4 n.a. 33/100 8/WB 3.45 iDX4-WB iS 1 3
SGS-Tho. DX4-V10HS 95 33/100 8/WB 3.45 iDX4-WB iS 1 3
Ti TI486DX4-G100GA 95 33/100 8/WB 3.45 iDX4-WB Cyrix 2 3
AMD Am5x86-P75 189 33/133 16/WB 3.45 iDX4-WB iS 1 13
Cyrix Cx5x86-100 130 33/100 16/WB 3.45 iDX4-WB Cyrix/iS 1 5
IBM 5x86-100c 130 33/100 16/WB 3.45 iDX4-WB Cyrix/iS 1 5
SGS-Thomson ST5x86 130 33/100 16/WB 3.45 iDX4-WB Cyrix/iS 1 5
iPODP (63 MHz) 429 25/63 16+16/WB 5 P24T iS 2 11
iPODP (83 Mhz) 489 33/83 16+16/WB 5 P24T iS 2 11
The prices for SGS-Thomson DX4-V10HS, Ti TI486DX4-G100GA, IBM 5x86-100c and
SGS-Thomson ST5x86 are only guesses, others the lowest street-prices in
germany. The iDX4-100 voltage of the newest chips is 3.3 V.
CPU special:
iPODPs: voltage reductor 5->3 V "on chip", fastest FPU in test.
iDX4ODPR: voltage reductor 5->3 V "on chip".
The mainboards, on which the CPUs were placed on:
Asus 486SP3G, Elitegroup EISA 486VE (5V only, 3V->5V adapter was used with
3V CPUs), Eitegroup UM8810, Soyo SY-4SAW2.
type on board 486SP3G EISA 486VE UM8810 SY-4SAW2
iDX4-100 1 1 1 1
iDX4-100WB 2 2 1 1
iDX4ODPR 1 1 - 3
AMD486DX4-100SV8B 1 1 1 1
AMD486DX4-120SV8B 4 5 6 6
Cy Cx486DX4-100GP - ? 7 7
Cy Cx486DX4-100GP4 2,8 - 1 8
SGS-Tho. DX4-V10HS 8 - 1 8
Ti TI486DX4-G100GA - 9 1 7
AMD Am5x86-P75 2 10 11 12
Cyrix Cx5x86-100 2,13 - 1 1
IBM 5x86-100c 2,13 - 1 1
SGS-Thomson ST5x86 - - 1 1
iPODP (63 MHz) 14 15 16 16
iPODP (83 Mhz) 14 15 16 16
-: Didn't boot in any jumpering
?: No data. Either forgotten or I missed it.....
1: No Problems.
2: L1 cache only WT
3: DX-4SL jumpering, L1-WT, L2-WB.
4: Only 100 MHz.
5: EISA bus on 1/5 CLK.
6: as DX4-100 _and_ overclocked to 40 MHz PCI-Bus. Be careful....
7: Jumpered as Cyrix M7
8: L2 cache only WT
9: Jumpered as iDX4-SL with internal WB without WB-signal support (5% perf.
loss)
10: 3->5 volt adapter must be jumpered to clock doubling to get x4
11: Jumpered to clock doubling, this gives x4.
12: Jumpers J21 & J22 short, manual is wrong.
13: Jumpering iDX4, without CLKMUL JP33.
14: Jumpering JP33: 2&3, 4&5.
15: Jumpering iDX.
16: Jumpering iP24T.
Now, BAPCo-95 benches, measured with Elitegroup UM8810, BIOS vers. 3.2F,
16 MB RAM, 256 L2-cache, Maxtor 7546AT HD, Elsa Winner 1000AVI for all
CPUs. The iDX4ODPR is missing (couldn't boot in UM8810), but the data
should be the same as for iDX4-100.
The iPentium-75 benches are from a Triton board with PB-cache.
type textpr. spreads. datab. des.gr. des.pr. DTP SYSm95 SYSint95
iDX4-100 188 186 274 211 185 190 209 205
iDX4-100WB 195 193 285 216 197 198 216 214
AMD486DX4-100SV8B 171 171 256 192 175 177 192 190
AMD486DX4-120SV8B 184 191 269 222 180 188 210 204
Cy Cx486DX4-100 157 156 243 243 159 161 173 175
SGS-Tho. DX4-V10HS 154 154 239 165 161 161 173 174
Ti TI486DX4-G100GA 154 153 240 165 160 160 173 174
AMD Am5x86-P75 WT 219 216 288 245 216 216 240 235
AMD Am5x86-P75 WB 227 228 320 255 224 225 251 246
Cyrix Cx5x86-100WT 202 200 293 226 189 203 224 218
Cyrix Cx5x86-100WB 206 204 298 229 203 209 228 225
IBM 5x86-100c WB 207 203 299 228 204 208 228 225
SGS-Thomson ST5x86 202 199 305 226 198 203 224 220
iPODP (63 MHz) 171 182 217 215 163 169 191 182
iPODP (83 Mhz) 237 251 309 291 222 231 264 253
iPentium-75 235 240 329 274 226 236 262 255
Summary: WB is good for 8-10% more speed. If you have a VLB-board the
AMD-120 is made for you, overclocking is always risky....
The AMD-133 is almost on P75 level, sometimes over. But there are problems
with the power-managment and the WB-cache on some boards.
Hope this is helping somewhere out there,
Uli
--
Dipl. Inf. Ulrich Teichert e-mail: kry...@netzservice.de
Stormweg 24 listening to: 45 Jahre (Chaos Z)
24539 Neumuenster, Germany Tod Ist Disco (Abwaerts)