Inside Abit - info over de BX6-II

Intro

De Taiwanners van Abit hebben in een korte tijd een dikke naam gevestigd als bakker van moederplaten en hebben daarbij een geheel eigen markt gecreeërd: die van de speciaal voor overklokkers ontwikkelde moederborden. En hoewel fanatieke Asus en AOpen hooligans de borden van Abit vaak willen omschrijven als goedkope low-quality crap, leuk voor de overklokkende hobbyist maar wegens een totaal gebrek aan stabiliteit ongeschikt voor ander gebruik, is Abit de absolute king of the hill als het om overklokken gaat.

Als je wilt overklokken is er maar één keus en dat is Abit. Kijk naar de Abit BH6, een ongekend populair moederbord dat over een aantal unieke features beschikt, zoals de 124MHz FSB en de SEL100/66# switch. Ook het voor ons overklokkers cruciale voltage tweaking en de Softmenu technologie, waarmee Abit als eerste fabrikant een jumperless moederbord presenteerde, zijn nieuwe ontwikkelingen die we aan onze helden bij Abit te danken hebben.

Met de komst van BX6-II, vergezeld van een vette feature list en een oplossing voor vrijwel alle gebreken van de BX6 en BH6, breken er wederom nieuwe tijden aan voor overklokkers. De BX6-II als godin der moederborden, een derde generatie BX bord waar de concurrentie geen antwoord op weet.

...en wat is er dan mooier om in direkt contact te kunnen treden met God himself? Wie regelmatig z'n surfplank bij AnandTech parkeert om even langs te wippen bij het message board had vandaag die kans, want ene Oskar, BX6-II co-designer bij Abit, keerde daar even terug op aarde om deel te nemen in een thread over Anand's BX6-II review, een review die trouwens zware nadruk legt op enkele (volgens Anand) zwakke punten van de BX6-II (waarover later meer).

De volgende pagina's zijn een compilatie van deze thread op het AnandTech BBS. Alle credits liggen daarom bij de originele schrijvers, ik heb er alleen wat in zitten cutten. Normaal gesproken gooi ik dit soort spul in de news engine, maar aangezien het daar na 2 dagen zo goed als weggerot is bouw ik er nu een artikel van, zodat-ie vereeuwigd blijft (tot onze database crasht ofzo )

Anand's kritiek op de BX6-II

Zoals gezegd bracht Anand in zijn review van de BX6-II een aantal kritieke punten van de BX6-II naar voren. Hieronder volgen Anand's opmerkingen, vervolgens weerlegd door Oskar:

A: The layout of the BX6 Revision 2 does reflect the same poor placement of the ATX power connector as the BH6, which leaves the power connector behind the slot-1 SEC connector, forcing you to extend the power supply cable over the processor in order to plug it in.

O: The location of ATX power connector is one kind of trade-off , we have our consideration such as making more stable switching circuit to supply vcore the cpu required.

A: ABIT's SEL66/100# configuration option no longer holds the unique value it once held in the market now that Intel has pretty much addressed the oversight that ABIT exploited.

Leaving the SoftMenu II setup without the unique advantage it once held over the competition, with AOpen already offering a jumperless setup on their AX6BC. The only reason a user would have for opting for the BX6 Revision 2 would be for the ability to adjust the core voltage of their CPU for overclocking purposes, a feature, that to a non-overclocker, can be quite useless.

O: SEL100/66# singnal does not only break the clock multiplier lock of the old Pentium II 350/400, it also effacts the level 2 cache speed. If you use 100 MHz FSB and set the SEL100/66# to low, you will find your winbench cpumark score to be faster

A: The stability of the BX6 Revision 2 is on the lower end of the spectrum when compared to the more well rounded solutions, such as AOpen's aforementioned AX6BC. The board's lack luster stability, a result of a general lack of high quality capacitors near critical components, brings down the overall quality of the board and the purchase value for those users that won't be overclocking to any great degree.

O: About the lack of high quality capacitors Anand referred to , I have to explain here. The capacitors you can find on our mainboards can be separated into three kinds --> Big electrolytical, small tantalum and small surface mounting ones, each of them has different kind of characteristics. High speed power decoupling required more small ones but not big ones. In fact, the big electrolytical ones cost less.

BX6 2.0 absolutely has no lack of high quality capacitor, we have different considerations to have such design in BX6 2.0 compared to BH6. Count the amount of big capacitors only is not the most important thing. For Example, if you check the capacitor for Vcore decoupling in BX6 2.0, you will find that all of them are bigger than the ones BH6 use. If fact , they are more expensive. We will not cost down the capacitor to sacrifice the stability

A: AnandTech's test system crashed an average amount of times during the stability tests, leaving the BX6 Revision 2 on-par with many lower class motherboards in terms of stability and separating it by a great distance from the competing AX6BC.

O: I am wondering why Anand feels the overclock stability of the BX6 2.0 is better than it's general stability . If one mainboard can not operate stable in a general situation, how can we expect it to be stable in overclocked situations. In our inner test, the stability of the BX6 2.0 is fully on par with the BH6.

A: In spite of ABIT's efforts to stay competitive, their newly introduced FSB settings are almost entirely useless. While the competing AX6BC made it up to the 143MHz FSB in

AnandTech's tests, an identically configured BX6 Revision 2.0 system had problems booting at frequencies greater than 124MHz. Attempting to run a system with an AGP card at any speed greater than 124MHz will probably be a problem in any case since most

AGP video cards have troubles operating at frequencies derived from such high FSB settings, so the inability for the BX6 Revision 2 to operate reliably at anything above 129MHz should degrade the value of the purchase any unless you will be using a PCI video card (which can run at 129MHz+ * 1/4).

O: I am also wondering why Anand feels that FSB's above 124 MHz are useless. In My personal tests of the BX6 2.0, the highest FSB the BX6 2.0 can achieve is 147 MHz with 8ns NEC SDRAM. The highest FSB on which the BX6 2.0 can operate stable is 143 MHz. I don't know what's the problem with Anand's BX6 2.0. I have to point out one thing: the highest FSB a mainboard can achieve will be limited by your SDRAM bank amount , this is a limitation of the Intel 440BX Chipset.

Q&A'tjes met Oskar

Over het probleem met de hoge AGP frequencies op bussnelheden boven de 124MHz zegt hij:

O: If you use a PCI display card, it's fully ok, right ? If you want to use an AGP card, there are still some cards that will do 133 * 2/3 = 88 MHz. At least our own i740 can perform at 88 MHz AGP clock even in 3D games (our i740 perform ok even at 100 MHz AGP clock in 2D applications, but failed in 3D applications. Some Matrox Mystique G200 cards also can perform ok at 88 MHz. I have once seen a post on a Taiwanese BBS telling that you can turn off G200's 2X AGP Mode, so it will perform ok at 100 MHz AGP clock.

If you want a safe solution, go with the diamond viper 550 TNT pci. I have tried some external circuit to implement 1/2 clock divider. It works, but this makes the agp card become a PCI device thus it can not cut-in Direct Memory Execute and Side Band Addressing. That made me give up the 1/2 clock divider solution. 1/2 agp clock divider is impossible, this is a BX chipset hardware limitation. The only safe way to use 143 MHz FSB is a PCI display card.

Meer vragen van bezoekers:

NFS3: Hey oskar, can you tell us anything about the TNT/BH6/Memory problem??

O: Our Taipei test center also tested for this problem about 1 month ago, but we couldn't duplicate the situation. In my personal opinion, there are three possible reasons for the hang-up:

1. The TNT display card core requires about 3.3V/2.5A for stable operation, this causes some ATX power supplies' 3.3V ripple to become large which causes SDRAM instability or even hang-ups.

2. If you run your TNT card at 1:1 AGP clock ratio, of course it will hang-up.

3. Some SDRAM module's SPD is filled in wrong parameter to cause this problem. If the point is NO.3, we are currently working on a new bios that fixes this problem.

Donatello: 1. Why can't we have some freqency that is higher than 83 but lower than 100mhz on Abit's BX based board??? (Intel's pressure (Overclocker knows) ? Lack of BX ability? or something else?)

O: Maybe it's possible in the future , but it is impossible right now because the lack of such kind of clock generator in the market.

D: If there are no Intel's pressure and lack of BX ability form question 1, can we just have update BIOS or SoftMEnu to set 8?mhz or 9?mhz???

O: The FSB frequencies we provide are depended on the clock generator we use. So it's impossible to have extra FSB via a BIOS Update.

D: Can Abit do FSB increment like VCore??? you know, 83mhz, 84mhz, 85mhz...

O: The same answer as no.1.

D: Can we do 1/4 frequency for PCI < 100mhz? I saw Anand's review, it says BX6 v2 can do 1/4 > 100mhz. For O/C's point of view, 300A should be out soon, 333 and 366 don't have good chance to do 100mhz, but, most of them can do well on 83mhz or higher, or maybe a bit lower than 100mhz! We all know doing PCI at 83/2 is not so safe in most of the case. Many people should have BH6. So, is this possible to do 1/4 or even 1/5 for PCI with Abit's BX board? If the answer is possible, can we just do a simple BIOS update?

O: If there is clock generator with frequencies between 83 ~ 100. It must feature a 1/3 PCI clock divider. The same, it's impossible to have extra FSB via BIOS Update.

D: About setting SEL100/66# to low, can people set it with Celeron 300A ~ 400 or just can't?

O: It's useless for current Celerons. Because current Celery's SEL100/66# is right now low by default. Speed up Level 2 cache may result in instable operation.

D: Will Abit do a hardware compatible result test and post it? Too many video cards, ram, and other PCI cards are outta in the market. This will help customers a lot. Especially, Video card and ram (ram issue with BH6).

O: Maybe it's possible, we currently have test reports but we don't make them publicl. I will talk to my boss about this.

D: Any 4X AGP motherboard news from Abit? I recently heard that SiS is going to have their SiS 300 2D/3D 128bits 4X AGP chip out to the market in April. BX doesn't support 4X, right? Any new chip coming out to support this baby? GX or JX?

O: 440JX ? you mean camino ? It's code name is i820. Every mainboard maker will have 440JX based mainboards when Intel launches the camino chipset.

D: how about AMD's 95mhz K6-2 cpu? What chip do other vendors use to generate that 95 mhz?

O: 95 MHz K6-2 clock generator can not be used in intel 440BX chipset mainboard.

Meer Q&A...

Nog wat over de stabiliteit van de BX6 II:

O: I am very confident about BX6 2.0's stability from my personal tests, in fact, the thing I care about most is the stability rather than overclock ability. Because I always think that if a system can not operate very stable in un-overclocked conditions, then it will not have any possibility to overclock.

Toasty666: I have a question for you all. Do you know perhaps how well the Socket 370 Celeron processors overclock? Iceman I believe said his worked @ 450 with 1.9V ! That's great. I'm right now in the market of picking up a 300a Slot 1, but the recent Intel 'scandal' has distracted me. Lots of people have also been saying that it is Stepping 5 (or 6) instead of the Slot 1 Stepping 0, resulting perhaps in better heat dissipation? Any input is helpful. Thanks.

O: I have tested with all the Socket 370 celeron 366 engineering samples I have. Ther's no one that can boot up at 550 MHz. I have no sense about 300A , but I think that the socket 370 CPU fan is smaller. This will cause a little difficulty dealing with heat dissipation. But that's just my personal point of view.

Xenocide: It has been said before, but thank you for taking the time to address these issues. You communicate effectively, and your English is more than sufficient! I have an Abit board now, and will have one in the future, since you truly seem to care about the people who use your products.

O: I think that making the customer satisfied is more important than just wanna make much much money from the customer. Only with proper benifit, proper service, one can make a company long-lived. It's impossible to make every customer satisfied , but I just want to do what I can do although the're still people not satisfied white the product. But making things better and better is always my goal. That's all.

Best wel koel dat zo'n gozer van Abit even langs komt om te posten op een openbaar forum zoals die van AnandTech, dat zouden ze vaker moeten doen (daarmee ontwikkelen ze ook veel meer begrip en vertrouwen bij de klant).

Anyway, lees vooral die thread op AnandTech door. Errug cool en er staat nog veel meer info. Voor Anand's BX6-II review moet je hier zijn en als je ook de mening van iemand anders wilt horen heeft Thresh's Firingsquad nog een review liggen.

Uiteraard staat ook ons forum open voor opmerkingen, vragen en andere stuff...tijd voor Oskar om ook eens hier langs te komen fietsen

Reacties (2)

ik vind dat Abit een kloten boort is en er zijn geen goeden boorden ja tjentech
is het besten boord en de 568 van tjentech

omdat je jan smit heet zal ik verder comentaar op jou mssg voor me houden