Toon posts:

A7V333 en Memtest 86 errors

Pagina: 1
Acties:

Verwijderd

Topicstarter
Heb met mijn verder uitstekend draaiende A7V333 van begin af aan - en dat met alle uitgebrachte biossen - het probleem dat ik mijn Corsair XMS3000 geheugen (2x 256MB)wel Cas2 kan laten draaien, maar niet met 1T command.
Zodra ik 1T command "enable" post de computer, doorloopt het bios en gaat het scherm, zowel in Win 98SE als in Win XP op zwart.
Voedingspanningen ok, temp ok, alle mogelijke geheugentimings geprobeerd, ....helaas.
Vanavond maar eens geprobeerd Memtest 86 te draaien.
Daarbij doet zich het (voor mij vreemde) voor:
FSB 133/33 en CPU/MEM op 1:1 geen errors in de (default) tests.
Zet ik het geheugen asynchroon 4:5 dan krijg ik t/m test 6 circa 64 errors en na 8 tests 120 errors.
Verhoog ik de fsb naar 144 en CPU/MEM 1:1 dan regent het fouten, iets in de orde van enkele duizenden.
Omdat ik geen ervaring heb met dit testprogramma is mijn vraag: Wat voor waarde moet ik hieraan hechten?

  • -=bas=-
  • Registratie: Oktober 2000
  • Laatst online: 17-07 05:15
Memtest werkt goed zover ik weet. Geen fouten gevonden betekent meestal wel dat het goed zit. En andersom ook :)

Als Memtest aangeeft dat het fout zit, dan kan je er vanuit gaan dat ie daar gelijk in heeft.

Senile! Senile Oekaki


Verwijderd

Topicstarter
Maar het merkwaardige is dus dat er bij een fsb van 133 en cpu/mem 1:1 geen fouten worden gevonden.
Die doen zich pas voor in asynchrone mode.
Blijft natuurlijk verder nog het probleem dat geheugen niet 1T command wil draaien, hetgeen zonder meer zou moeten kunnen.

  • CyberTijn
  • Registratie: Juni 2001
  • Laatst online: 04-08 16:00
Ik begin de laatste tijd echt te twijfelen aan Memtest86. Met de opkomst van DDR333 verkoop ik vrij veel MSI KT3 borden met 333 geheugen (werk in computerwinkel).

Echter, ik heb zelden een werkende PC2700 module gezien, van zowel brand/brand (samsung) als brand/3rd. Al die modules geven errors in memtest86.

Verwijderd

Topicstarter
Heb nu 1 van de 2 modules er uitgetrokken, t.w. de "verdachte". En....jawel geen memtest errors meer,
1T command werkt, memscore hoger, 3Dmark2001 SE loopt niet meer vast.
Dus maar 1 module omruilen bij leverancier.

  • Mr Magic
  • Registratie: Juni 1999
  • Laatst online: 15:00
Ik las in deze review het volgende interessante stukje informatie m.b.t. het voltage dat de DIMM's krijgen op een A7V333:
DDR Voltages

When it comes to running voltages out of spec, hardly any manufacturer has the kind of track record as ASUS. From the P5A to the A7M266, the road has literally been paved with burned DIMMs and the A7V333 threatens to continue this tradition.

The layout of the VDDR jumpers has been carried over from the A7V266-E, that is jumpers JP1 and JP2, only the values have changed. By default, the A7V333 supplies a whopping 2.78-2.85 VDD and VDDQ (DDR core and I/O voltage). The variation accounts for the differences in measurement on several boards. Changing the jumper configuration allows settings between 2.59 and 3.06V. The detailed settings are:

ii 2.59 - 2.63V (same with one or both jumpers removed completely)
i! 2.78 - 2.85V (default setting, out of spec according to JEDEC guidelines)
!i 2.88 - 2.96V
!! 2.95 - 3.06V

The latter two settings are almost guaranteed to destroy the memory modules, maybe not today but after a few weeks. It is not clear what the purpose of the high voltage settings is, however, it is only a matter of time until this will become a liability issue for ASUS, one way or another. Keep in mind that the settings are not documented and the high settings are not meant for "public consumption" by ASUS either.

The interesting thing about these settings is that all current high-end DDR chips are using internal voltage regulators to reduce the core voltage to 1.8V which is necessary to run at high frequency. The only parts of the chips that actually see the increased voltage are input/ output buffers and, granted that those will wiggle a bit faster at higher voltages, the net effect is counteracted by the fact that the internal voltage regulators produce a lot of heat that will overall slow down the chips. In other words, a sensitive approach would be to leave VDD alone but increase VDDQ. Admittlely, there are some older DIMMs that will run at 2.5V internally but those will hardly be used in PC2700 mode of operation which is the only scenario where the massive overvoltage would be of benefit.

The recommendation in this case is, similar as what we suggested for the A7M266: simply take off both JP1 and JP2 to bring the voltages back to where they belong. Side effects are that the failure rate, that is, stuck bits and system crashes in memory-intensive applications go down quite dramatically. In addition, we saw up to 3 % system performance increase in Expendable at the lower voltage. It is not completely clear what causes this performance increase but any electronics will run faster and cleaner when they run cooler.

Verwijderd

Topicstarter
Ja dat had ik ook al gelezen, op zich interressant want iedereen denkt dat een hogere dimmspanning tot een betere performance leidt.
Pagina: 1