VIA Insides: Wenchi Chen speaks up

12.04.2002 von NICO ERNST 
At CeBIT 2002 we took the chance to talk to the VIA-CEO once more about bugs in chipsets, memory technology and VIAs CPU strategy. In the end, Chen promised to have his southbridges redesigned.

Chen arrived in Hannover two days late, missing the VIA press conference. As soon as he touched down, he disappeared in meetings again. Only on Sunday morning we had a chance to meet with Chen, who surprisingly seemed as relaxed and entertaining as ever.

tecCHANNEL: You did not miss your own press conference at CeBIT in years. What kept you up?

Chen: Actually, I miss my press conferences all the time. That depends on my schedule. Also, we have a stronger (PR, Ed.) team right now, so they got to do more work.

tecCHANNEL: Many people at the show rumored that you were negotiating with ST Microelectronics to buy their Kyro technology.

Chen: Oh, I won't comment on anything like that...

tecCHANNEL: Will you comment at a later point of time during the show?

Chen: Not until the deal is settled.

"We did not get this kind of input"

tecCHANNEL: OK. When we met here last year, we were discussing a bug in your 686B southbridges, resulting in data losses. Then tecCHANNEL found out that there is something else wrong with the way you implement PCI.

Chen: Right.

tecCHANNEL: Last year you said that you will put in place new customer service practices, faster response times - how could it happen that we just found another bug?

Chen: This one is related to performance. It is not exactly related to product compatibility issues. We have been shipping this product and have a lot of people using it - we didn't get this kind of input. That's something we continue to learn from this business: There's always something new. We just need to continue to work harder. We need to continue to improve the quality, performance, all these kinds of things for our products.

tecCHANNEL: I'm sorry, Mr. Chen, but I have to disagree, it's not all about performance. We have been trying to communicate that to VIA for four months now: It's also a stability and a compatibility issue. When we published the story we had lots of feedback from readers, not only using RAID cards, but also using sound cards, where the sound was crackled up as soon as you had a constant stream of data. We then did an analysis of almost every other chipset by Intel, SiS, ALi, and your line of products is the only one that has this problem. We think, there is something wrong with the way you implement the PCI on the Southbridge.

Chen: We can argue if that's (a problem of, Ed.) performance, or not exactly performance. No matter what: I think that's something we need to deal with, and we need to work on. I don't think we need to argue, one way or the other, that doesn't really matter. In the end, the more important thing is: How to deliver a better product?

"The improvement is in our new design"

tecCHANNEL: Are you planning to have the southbridges reworked? I think we haven't seen a major update there for the last 18 months, it's still the 686 design for most of your parts.

Chen: In this business there are many issues. When we got this input, we had our team analyze it and looking at what kind of effect it will have. But then we have many customers using our products. It's not easy for any major customer to check a new product, it takes a lot of validation work. But: The improvement is in our new design. In our future products, we will be thinking of this.

tecCHANNEL: And we're not talking about just another patch, are we?

Chen: No, no, no....

tecCHANNEL: We could sit here and talk about more exciting stuff if this was out of the world - at your press conference, VIA said that they want to make more than 50 per cent of their revenues with non-chipset products by the end of 2003. Why?

Chen: For a couple of reasons. Number one: We want to grow. We already are a really substantial chipset player. So, how are we going to grow, if chipsets are always more then 50 per cent of our business?

Number two: We do have a new exciting product lineup. We recently announced our "Project Canaan". We set off starting with five business units, and will add more business units. Besides the chipset business, we will also have 3D graphics, embedded CPUs - that's a very exciting product line! We have been attracting a lot of interest, even though the CPU business has not taken of as we would like to see it. In the past three years it has become very severe.

"Intel and AMD did a fantastic job"

tecCHANNEL: It's a battle...

Chen: In the past three years both Intel and AMD did a fantastic job on the CPU side. That has definitely been impacting our objective to deliver. But then, I think we found a good direction: Fanless CPUs. And then also a very big one is our CD-ROM, DVD and CD-RW product line. So overall, we see a lot of new opportunities.

tecCHANNEL: Are there any plans to do more on the communications side?

Chen: Any business that is big enough, we'll be interested in. We are in the component manufacturing and pride ourselves with high quality, high performance and low cost solutions.

tecCHANNEL: You are moving very fast with shipping the KT333 and already announcing KT400 and other DDR-400 products. Don't you think that's confusing for the customer?

Chen: I think that's the general problem we are facing. If you look at it for a long time: We moved our products according to the CPUs too much. When the CPUs move slow, we can afford to move relatively slow. Right now with the CPU moving so fast, in fact our chipsets and their memory bandwidth are still matching the CPU bandwidth. So, we need to continue to match up our chipsets to get some expansion in performance. Of course, this is pushing the whole industry - it's a challenge. We didn't realize it is going to be this fast.

"If DDR-400 won't work at all - that's fine"

tecCHANNEL: Intels vice president Mike Splinter just said the other day that they will move their 3 GHz CPUs into this year....

Chen: They told us much earlier on when they want to do this. So we need a fundamental analysis regarding the bandwidth requirement to deliver faster products. We will continue to be the leader in that space.

tecCHANNEL: That's challenging the memory vendors as well. Infineon and Micron said at platform conference that rather want to move to DDR-II directly and not do DDR-400 at all because the signal integrity and the noise level were just too challenging.

Chen: We are doing both. In the PC business you should never say never. If that was the case, PC133 would have never happened. We definitely agree that DDR-400 is very difficult to do. But on the other hand, DDR-II won't happen any time soon. If DDR-400 won't work at all, that's fine, but so far, we are already able to show something with DDR-400. So we know these DRAM makers can do wonderful miracles all the time (laughs).

tecCHANNEL: Let's hope they can create a few more miracles....

Chen: But we need to be ready, we need to prepare. As you can see, VIA right now is on the drivers seat, driving performance - together, for the industry. So, we need to prepare these things, we need to put in a lot of validation requirements. We also want to include the DRAM makers, because we want to make sure that the market sees the best performance, but still with a high quality product.

"Everybody who has Rambus wants to dump it"

tecCHANNEL: Right now, in the market, Rambus modules are available at the same price as high quality DDR modules. Do you think that in the future Rambus will be a little more interesting for the desktop?

Chen: You are talking about the market price - we are looking at the cost. The market price is variable, it's always up and down. Right now, of course, Rambus can be cheap, because everybody who has Rambus wants to dump it. So, they better dump it fast. So, it's not a good comparison. We were always interested in Rambus as a good research project.

tecCHANNEL: You were hinting before that VIAs CPUs will stay on the "Cool Computing" side for specific applications. Any design wins for the Eden platform yet?

Chen: Oh, lots of them! But we want to make sure it's not only us saying that, we want to see the real interest. When we look at the PC business now, the excitement is not as high anymore. When we look at all the major PC OEMs, they don't do much R&D anymore. They're box movers. With the new Eden platform or the Information PC there's lots of new excitement, lots of new companies, new startups.

tecCHANNEL: Of course, the last question has to be: How is your conversation with Intel going on the P4-bus license?

Chen: Oh, you know I won't be able to comment on that. If I can comment, we'll say it.

tecCHANNEL: Thank you. (nie)