3DNews Vendor Reference English Resource - All you need to know about your products!

Processors

Let's first start with the overall situation on the market and then go on to dealing with particulars. As per In-Stat/MDR, AMD first went up from 16 to 20 per cent of the x86 processor market share for the year 2002 and then reverted to its previous 16%. At the same time, Transmeta and VIA had the overall market share as low as 1%, so things for them are going to progress in the same course as before further on.

This data fully conforms to the fresher evidence provided by Mercury Research. As per the analysis, the market share for x86 taken by AMD in the second quarter of the year amounts to 15.7%, which by the way is 0.1% greater than a year ago. Intel's share is 82.5%, which is 0.3% lower than for the same period last year. The aggregate share of Transmeta/VIA is 1.8%, which is 0.1% higher than for the same period last year. In the interpretation of Mercury Research though, the situation for AMD looks more optimistic. Maybe it's due to that recently AMD finally has got down to making practical moves to catch up with Intel's Pentium4. But that applies more to the server market rather than the desktop processor market. As regards the mobile processors, AMD has nothing to oppose the Pentium M. Therefore, its market share as per IDC in the second quarter dropped by 2.5% having shrunk to 10.6% versus 88.1% for Intel.

Despite the 16% market share of the x86 processors, this is by no means a great success for AMD - the company has managed to hold the second place for all these years, while some players have vanished out of the market. Another thing that helped AMD keep afloat is the Athlon processor released four years ago and which proved a real breakthrough for the company. With minimum amendments added, this core is anyway able to compete Pentium 4 albeit originally meant for competing Pentium III.

Moreover, it is still progressing: in August there came the news that AMD suddenly decided to reanimate the Duron brand through releasing processors on the 0.13 mk Thoroughbred core with merely 64K of L2 cache (the result was dubbed Applebred). What is indicative, they were marked following their true clock speeds - from 1.4 to 1.8 GHz. It is promised that at performance even the 1.4 GHz version will be superior to Celeron 2.0 GHz (seems to be true at specific tasks for the cache of that size), and the prices are projected to be at the minimum level - between $32 to $47. Somewhere in our country such processors are already available on sale.

This applies more to the "newly-baked" Opteron processors - these are growing up before our eyes and gaining much success. Early in August, when the single-processor Opteron versions 140, 142 and 144 were already there, AMD announced its new processor for the dual-processor systems - Opteron 246. Three weeks afterwards, later in the month it appeared in retail shelves! As regards the success, it's IBM which contributed most through presenting Opteron-based servers in the previous month. In August, IBM declared it would produce workstations on its base in September onwards. Of course, IBM knows it better - it is now vitally interested in the success of AMD processors, and as a manufacturer the company acquires these processors at the cheapest possible prices close to the prime cost. Its further plans are to present Opteron-based and Fujitsu-Siemens workstations.

The past month was also notable for an ongoing demand for Opteron among manufacturers/customers of supercomputers. The well-known consumer, US National Research Laboratory in Los Alamos, is planning to use these processors for creating two large Linux-based clusters with the total amount of processors to be over 3300.

Things with 64-bit processors are going much sadder for AMD - August was not remarkable at all but for rumors, guesses and hypotheses. On the official side, of note is merely a presentation of new Opteron and Athlon 64 logos. And there's nothing of a surprise - the official date for Athlon 64 announcement is still 23 September, so we can't expect anything more.

There is some information like persistent rumors saying that things with the new processor for AMD are not going well enough (which is not a sudden news though), so the new Athlon 64 might be released with downgraded ratings in order to effectively compete Prescott at least at ratings. Therefore, the 2 GHz version (priced at $399) will offer the 3200+ rating - not 3400+ as expected earlier. For now, it doesn't make sense messing around with various sockets for the new processor - 754, 939, 940 and Athlon 64 FX (Socket 940 Athlon 64 appears to be an Opteron with the L2 cache cut-down to 512K) until we know it all from the primary source.

Another interesting aspect related to the release of the new AMD processor is that in future some versions with the 64-bit addressing disabled, sold as Athlon XP but running on the same motherboards as Athlon 64 may appear. The reason may be they are all packaged in the same 754-pin case. Actually, why not? - the unified infrastructure is a pretty handy thing in all ways. In fact, if that does happen, it's not going to be earlier than late next year.

The Au1100, a highly integrated solution and a response to the XScale, has been produced for quite a while in the 333 and 400 MHz makes, and AMD has been offering the developer's kit for some time - AMD Alchemy Solutions Mobile Handheld Reference Design Kit and OpenPDA Development Studio based on Linux. At LinuxWorld in August, quite smart PDA prototypes based on the Au1100 were presented. However, things are not going further than that anyway.

The 'high-life chronicle' is bringing us back to the beginning of this article, to the part mentioning those competitor players that gone for good. AMD decided to purchase the remains of one of those players in August. Some time ago, National sold Cyrix to VIA but retained the division in charge of the x86 Geode processor featuring integrated north and south bridges. That's just the division AMD acquired - to all appearances, that seemed to be the reason VIA gained some relative success with its integrated Eden.

Also, AMD declared in August it would provide technical support for Eden for the coming 5 years thus trying to attract more users. The company is not relying much upon VIA's assurances and presented its own another product - Eden ESP7000. To date, this is the fastest representative of the family which along with its 733 MHz clock speeds, including the integrated hardware random-number generator, resembles its predecessors. The power consumption of VIA's new offspring isn't much of a record: 6W for 733 MHz isn't so little.

Transmeta's new product - Crusoe TM8000 renamed to Efficeon - doesn't inspire much hope. Indeed, the company asserts that at performance the product is superior to current Crusoe by 50-80%. Actually, its performance wasn't so high even compared to the mobile Pentium III. But this time it has to compete with Pentium M. Its current 1 MHz clock speed is no longer impressive even with the integrated Hypertransport 400 MHz bus controllers, DDR400 and AGP 4X onboard. Anyway, let's wait and see, - we've been waiting for it for several months by now.

The price is an important factor - for Pentium M, it has gone down essentially and will drop further. By now, the situation with price reduction for this product line anticipated in October has been lastly defined: the drop is expected to be within 13 to 34%, therefore, the 1.7 GHz model will turn cheaper from its current USD637 to USD423. At the same time, prices for the mobile Pentium 4 will drop by 14-35% in no earlier than late October, just when prices for their mobile analogs go down.

August brought more joy to those into regular inexpensive PCs: at the end of the month, prices for higher-end Celerons were brought down by 6-14%, and by 2-10% for mid-end Athlon XP. Therefore, the officially announced wholesale price for Celeron 2.2 and Athlon XP2100+ is currently set to $69.

As regards the main Intel's processor Pentium 4, the situation with it is shaping up in a quite curious way: as distributors say, the company is intentionally reducing the volumes of shipments for the currently most popular FSB533 processors in order to make the end-users migrate more actively to FSB800 Pentium 4 processors and thus to i865-based motherboards. This autumn should bring quite a pleasing effect: price reduction for Pentium 4 and accompanied by the launch of mass sales for i848P-based motherboards will indeed create all the grounds for the consumer to migrate to the FSB800.

As usual, Intel is manipulating the market the way it wants in order to improve the financial results, and the company is doing a good job of that. This time again: Intel has already increased the forecasted sales volumes in the current quarter from $6.9-7.5 bln to $7.3-7.8 bln. As they say, because of the better than expected demand for the processors. Why on earth are these distributors complaining?...

The company is also fully optimistic about the future and is busy arranging the 90 nm production at the D1C pilot plant releasing Prescott and Dothan samples. In the fourth quarter, the Fab 11X factory in New Mexico will start a mass production of 90 nm chips. There are problems not yet solved of course, not only with the production. Intel engineers are currently working at reducing the heat emission (103 W at 3.6 GHz!). Prescott is famous for its projected short life - in autumn we are expecting 3.2-3.4 GHz processors, which is going to end up reaching the 3.8 GHz bar probably in the second quarter next year. It is assumed that in the fourth quarter of the next year the processor will be replaced with the FSB1066 Tejas at 4 GHz clock speeds and higher.

Anyway, this is nothing compared to the mobile Dothan completely deferred until the first quarter of the next year instead of the fourth this year. What happened? The yield ratio is too low with the heat emission too high - that is much more critical for mobile systems rather than desktops. As rumors have it, the heat emission in >1.8 GHz Pentium M turns out to be 5-10W greater than the pre-planned 30W, i.e. by 15-30%. Clearly this issue can't be solved through installing a more powerful cooler, the way Intel does with desktop processors.

This question should be addressed to the developers. The coming IDF is already at hand, and Prescott with Dothan will be at the center of most lively discussions, along with other further products like 90 nm Xeon (Nocona) to be released already in the fourth quarter, and its multiprocessor brother Potomac scheduled for the second half of the next year and so on.. Up to the 64-bit Tanglewood featuring 16(!) physical cores all made on a single chip planned for release in 2006-2007.

Motherboards

The hero of the month is evident - Intel or i848P, depending on the viewpoint. In any case, this won't change the sense much: it's just this chipset that gained maximum popularity among motherboard manufacturers in August. To a big dissatisfaction of memory manufacturers, the i865 has not proved to be the catalyst for the mass migration to dual-channel DDR400 configurations despite the so much hope built upon it. The users voted for more optimal processors in terms of price-performance - those featuring FSB400/533 and motherboards based on the i845 line.

Judging by the sales volumes, Intel has to hastily make use of the i865 cut-down version featuring the single-channel DDR400, which is just what the average end user wants and who again prefers evolution rather than revolution. Although the chipset hasn't been officially announced, August brought a flood of reviews of motherboards built on its base. Of course, the leaders MSI 848P Neo and ASUS P4P800S were ahead of the others. Then others followed suit, so by the end of the month the i848P was presented in the product lines of the most significant motherboard manufacturers. Some of the products are already appearing in the retail.

However, announcements is one thing, but the real sales is quite different. Although the i848P does look good, its competitors rank much better. The official price of the chipset is set to $29, and large manufacturers purchase it at $27-28. VIA and SiS offer their produce featuring similar specifications at almost half the price - the PT800 is priced at $15 and this time has no contradictory evidence against use. This immediately brought the results - e.g. the P4VTB made by Biostar appeared in the retail in Japan at merely $75 which is 15% cheaper than the cheapest i848P-based boards.

All in all, the option looks almost ideal, if VIA hadn't spoiled everything through delays with PT800 shipments. The cause - incomplete tests of the Serial ATA interface included in the south bridge VT8237 - appears to be quite dual. On the one hand, it's a good sign that VIA is not planning to promote a raw product to sales, on the other hand - this is not at all a reason for delays. The problem is not so big though, and as per the totals of September we will see a long list of announced motherboards based on this chipset.

As was already said, there aren't legal problems any more, so VIA merged the VPSD division with the processor division thus restricting the activity in the sphere of motherboards to the EPIA platform which is still on the up. For this year, the company intends to increase the C3 processor sales volumes by several times as compared to the previous year, and as another proof the company presented a new mini-ITX motherboard - EPIS CL aimed at the market of light servers.

VIA has always been successful on the market of chipsets for Athlon, and boards based on the K8T800 for Athlon 64 and Opteron have been actively announced for several months. August was no exception - as the dates of releasing the processor are coming closer, the activity in the sector is growing, and in the past month we also saw about 5 new or pseudo-new models announced. All this is warmed up by rumors on a new chipset for Athlon XP. The processor will still be popular for quite a long time. The nForce2 is increasingly on the up and it's high time something is released as a response to it.

As rumors have it, the KT880 - a dual-channel DDR400 chipset with support for FSB400 Athlon XP might be such a response which is going to see the light somewhere in mid-autumn. There is no evidence of how its multimedia functionalities will match the nForce2, but such a high level is often not necessary, and the VT8237 as a south bridge looks impressive enough.

And it seems like the competitor has stopped its advancing - as they say, new versions of nForce south bridges will appear only in the next quarter - not in this as was planned earlier. So the boards using the MCP-S and MCP-S1000 will appear on the market somewhere in September at best. Let alone the nForce5, a hypothetical chipset of NVIDIA for Pentium 4: as per rumors, the company has finally signed a license agreement with Intel, and now the development of the chipset is progressing at full sway. Among the details, the following features like support for the DDR2, Socket T and PCI Express are mentioned, so we are unlikely to see even an announcement of the chipset this year. On the whole, the information looks plausible: it's better to pay high royalties from something rather than no to pay them at all thus not earning nothing. But on the other hand, ignoring the market of Pentium 4 chipsets because of high royalty rates would bring NVIDIA much more losses than profits. VIA has finally come to understand this.

NVIDIA still has to rely upon the Athlon: nForce2 Ultra 400 is still used by motherboard manufacturers and is increasingly moving to the Low-end sector - in August, Acorp released a motherboard based on the chipset at $66 recommended price, as well as for Athlon 64 and Opteron. Soltek issued an announcement of an nForce3-based motherboard at the end of the month (together with other boards built on the VIA K8T800 chipset).

We seem to have forgotten about SiS, don't we? It's SiS who is able to win a piece of the market from NVIDIA - to all appearances, Microsoft will prefer NVIDIA graphics in the XBox2 rather than ATI. And since ATI is not doing well enough with south bridges, then a respective south bridge chip from SiS could be used in the XBox. At that, SiS has traditionally been a winner. Well, we've got to wait until the year 2005 and see it then.

For now, SiS is busy delivering its dual-channel SIS655FX to motherboard manufacturers who are actively using it (this primarily applies to Asus of course). Among the other products is the SiS741 chipset as a reminder of long-forgotten area of Athlon chipsets. At last, the company decided to upgrade its integrated offering SiS740 for this processor and added support for FSB400/DDR400 processors, accelerated video core, SerialATA and RAID ports etc. On the whole, despite the modest addition of the '1', the product proved to be up to the mark. Interestingly, the boards are expected to arrive already in September onwards. Looks like this autumn is going to be quite hot for Athlon chipsets.

Memory

Despite the numerous pricing forecasts of July, prices for memory didn't grow up. Even contrary to that - a complete lack of demand during the month, despite the aspiration of manufacturers, caused the price bar go down to the minimum level. But at the end of the month the price drop stopped and some stabilization occurred. Maybe as per the totals of September we'll see some growth due to the following two reasons.

First, the situation on that market looks a bit strange when DDR400 modules are not in sufficient demand actually, so the prices for them froze, but the prices for DDR333 were growing until they equaled those for DDR400. That's it. DDR333 can't cost more than DDR400, but prices for the latter are not going up because nobody buys these modules. Now that the i848P and PT800 are released, things with demand for DDR400 should get better essentially. Secondly, the demand should grow anyway - it's the season, and summer is over.

The situation is the same on the market of memory modules: in fashion are overclocking wise modules, but all these 422,466 and 500 MHz are so boring that the user perceives all of them like something customary, so it's high time manufacturers invented something brand new. The Twister line of TwinMOS offering speeds up to 500 MHz are no longer topical. DDR550 chips and "PC4250" modules on their base are already the latest craze of fashion, sort of those Corsair is planning to present in the fourth quarter. This also applies to dual-channel DDR533/PC4200 presented by GeIL.

The hero of the 'high-life chronicles' is definitely Kingston who managed to increase the sales volumes for the modules from 80 thousand to 2.5 mln. Moreover, in the nearest future the company is planning to increase its production capacities either through adding extra production lines at its factory in China or building another one nearby. Clearly such volumes are tasty morsels for chip manufacturers striving to become suppliers of Kingston produce. That's where the main intrigue of the month was hidden.

Naya, until recent times being the main supplier of Kingston, decided to launch production of modules under own trademark, which affected the relationships between the companies in a most negative way: one hardly ever likes it when your supplier starts competing against you. As a result, Kingston changed the course of activities. That concerned Powerchip, Infineon and Micron, so Nanya lost billions of dollars in terms of turnover. No one knows if the production of own modules is worth it.

During the coming autumn, Elpida is going to be just another supplier of Kingston. What's most interesting is that Kingston will pay for that - through purchasing shares at $50 mln, which will allow Elpida increase the yield rates and expand the production of 0.11 mk produce. The share of supplies in the total production volumes of the company will be as much as 5-10%. This money will be of use for direct expansion of production capacities - the company leased a 200-mm wafer factory from NEC, one of the founders.

This amount will somehow help the company build up the share on the DRAM market, which dropped to indecently low 3.6% as per Dataquest, while Nanya's share for the period has turned 4.4% even though the company was very little known mere three years ago. With these parents and such a history (Japanese used to hold 80% of the world's DRAM market), the company might have behaved better.

The Big Three players on the market remained unchanged - Samsung is still controlling one third of the market, at 31%, Micron takes the second place at 18.9%, and the third place is taken by Infineon at 15.4% of the market. In fact, Hynix is closely behind, and the company's share on the market for the past quarter has somehow gone up - to 13.7% (by as high as 1%). However, in this quarter the company's affairs are unlikely to go so well - the adverse effect of duties is there, then the company's plans to sell three memory production lines at its factory plus two enterprises operating in the same field. 

Hynix's DDR II modules

Video

In terms of products, the month proved to be a real summer month - without sensations but full of interesting events mainly in the sphere of high-life chronicles. The major topic is of course the potential strengthening of ATI's positions. It looks like Asus and Elitegroup, two strongest market players, are joining the company. As per rumors, the former is to announce the launch of new cards built on ATI chips at Computex. What is remarkable, it has been cooperating with ATI for some time already through manufacturing motherboards based on ATI chipsets under own brand name and as an OEM is making video card for the clients. This time what is meant is the release of Asus-branded video cards, and this looks quite plausible. Today nobody can ignore the growing popularity of ATI chips. The main intrigue is not about "will be - won't be", but in that whether Asus continues manufacturing video cards based on NVIDIA chips as well. Up till now, the jealous company has deprived anyone who contacts ATI of the privilege.

The second new is about ECS' decision to tackle the manufacture of video cards based on ATI chips in real earnest. Nothing prevents this company from doing it, since the company hasn't been noticed at producing video cards on NVIDIA chips under own brand or video cards on the whole (except the produce built on SiS chips which can hardly be regarded as video cards so far), so the company has free hands at that. The company intends to start manufacturing a full spectrum of products - from Radeon 7000 up to "High-end solutions", so by the end of the autumn we may see an influx of cheap produce of the kind.

Finally, the third news comes from the same field - it's Gigabyte decision to acquire ELSA. At least, the negotiations on that matter have been under way successfully for quite a while. This would allow Gigabyte to expel FIC from the sector of video cards for ELSA and acquire a well-promoted brand which is by the way associated with high-end solutions. In August, ELSA announced two new lines of professional video cards based on ATI chips - FireGL X2 and T2.

A good month for ATI, wasn't it? Microsoft announced for sure that ATI chip, not NVIDIA, would be definitely used in the Xbox2, since the company's technical vision matches the progress trend of Xbox. Judging by NVIDIA'a experience, this may result in serious losses rather than profits, if Microsoft fails to adequately estimate the demand. Anyway, analysts say that in 2005 (planned terms for releasing the Xbox2) ATI will gain extra $25-35 mln profits, but NVIDIA might lose about $250-300 which could have been gained under a different course of events.

In the year 2005: at least R500 will be used in the Xbox2. For now, the press is lively discussing the rumors of the R360 core which might come up this autumn in the Radeon 9900. The prospective specifications are these: 0.15 mk (!) process technology, core speed 420 MHz, memory clock speeds merely 750 MHz. The performance is middling, I must admit. The only thing to hope for is the not too high a price. Somewhere around this time, there might appear the budget solution RV360 which appears to be at least a little better than the R360. So for now it doesn't make sense taking it all for granted - in September, things should finally clear up.

In any case, ATI is advancing very well, and it's no surprise that in the third quarter the company is planning to increase the volumes of orders placed with TSMC and UMC - from 70 to 100 thousand wafers per quarter, thus becoming the largest client of TSMC (currently, this title is still in NVIDIA's possession). Despite that, NVIDIA is still the largest seller of graphic chips and holds 27% of the overall market as per the totals of the second quarter. On the other hand, in the first quarter the figure was 31%, which 20% for ATI then and 21% - now. The gap is shrinking just before our eyes. In fact, things for NVIDIA on the desktop chips sector are not progressing that sadly for NVIDIA - the company still holds 64% versus 28% by ATI, and for the second quarter these shares haven't changed.

In money terms, the operation results for NVIDIA in the second quarter look quite decent, although not bright: $24.2 mln the net profits against $459.8 mln turnover. By the results of the first six months, the turnover dropped as compared to the same period of the previous year by 16% and reached $864.8 mln. Actually, NVIDIA is not to blame for that - recently, the market of integrated produce has been increasingly gaining momentum, which the sales volumes of video cards are continuously going down. For the first half of the year, none of the largest video cards manufacturer has been able to achieve the originally posed goals.

So, at that NVIDIA and ATI are in the same boat, and it's no surprise that competitors had to undergo coordination of their activities on the market. In August, there were rumors that both companies entered a classical collusion - an agreement on fixing prices or at least on loosening the heat of competitive struggle. In theory, such a breach of the antimonopoly law is highly probable to have taken place: the cost of development for new chips is increasingly going up, and the deadlines for releasing new versions are continuously shrinking, so it might end up going bankrupt if the current trends persist.

NVIDIA is still having immense plans which will definitely require huge investments. All that is going on with rumors saying the company is holding negotiations with Sony on the matter of using its chip in Playstation 3 (sort of a revenge for Microsoft's treachery, but that's unlikely because NVIDIA can hardly get something from that - the universal Cell needs to be promoted, doesn't it?). The month was also notable for MediaQ acquisition, a company involved in developing graphic and multimedia technologies for various mobile devices targeted at PDA and mobile phones. Finally, the plans for developing the core business with the monstrous NV40 whose presentation is expected at Comdex (very unlikely, though) and which is made up of 300 (?!) mln transistors. Well, seems like a graphic Itanium in your desktop PC, doesn't it?

For now, while there aren't any new chips, the company appeared in a quite unusual way: the main products built on NVIDIA chips announced in August became the devices that allow turning a PC into a digital VCR. These are the video-in and video-out, the TV-tuner; specialized software and quite often a remote control. In August, there were at least four such devices based on GeForce FX 5200 and 5600. It's also worth noting the emergence of cards on the base of GeForce FX 5900 featuring reduced chip and memory clock speeds. This of course applies to the price either. Originally, it was too high, especially in view of how fast ATI reduced prices for its Radeon 9800SE.

Anyway, however low the prices are set, it's a too long way to get Xabre for both NVIDIA and ATI. A mere trifle remains - to make the company release its long-promised chips. To get that, almost no time is left - another promised date for the release of XG40, XG41 and XG42 is September. Yes, that's it - XG-something, since SiS deprived its daughter enterprise of the right to use Xabre brand. It's a real pity, because the brand was well-promoted. Well, let's leave the brand - let the chips be first released, and then we'll see..

Data storage

The month was fairly calm with not so much information - in fact, this section offers less news than other sections do. Against this background, the customary amount of news appears to look unusually large. Of most interesting high-life chronicle news is evidently Hitachi's decision to transfer its HDD production facilities over to China - now 40- and 80-GB Deskstar 7K250 hard disks will be produced by ExcelStor, a daughter enterprise of the Chinese multi-profile concern Great Wall Technology Limited. At the same time, the company has got the right to produce their analogs under own trademark - ExcelStor. By the 2006, the vast majority of hard disks produced under Hitachi trademark won't be manufactured in Japan any more, but in China.

This is the "high-life chronicle", but none of the hard disk manufacturers stood out with new products in August. Seagate presented a family of CE hard disks aimed again at the consumer electronics market, with the logics optimized for simultaneous multimedia streams. Along with that, a 2.5" server hard disk for 1U servers featuring support for the Serial Attached SCSI interface was finally presented.. Maxtor presented a new generation of external hard disks OneTouch featuring a new design, increased disk capacities (120-300 GB) and a set of servicing software.

In fact, the most interesting hard disks in August were those called "little ones", not full-sized. For instance, Fuji demonstrated prototypes of HDD plates of the 0.8" form factor, which offer quite acceptable capacities - up to 6 GB per plate and recording density of 80 GB per square inch. It offers the customary maximum of five plates, and the tiny HDD may have disk capacities up to 30 GB, which is more than enough for using it as a media even in digital cameras. And Fujitsu is even targeting at the market of mobile phones.

Well, there aren't any visible signs of wide use of the 1" Microdrive whose 4GB samples by the way started being shipped by Hitachi. Maybe the speed is not enough - but the company did increase it by 70%, or maybe the price ($500) is a bit too high? Anyway, Hitachi didn't take much trouble at improving it in terms of price, and all efforts seemed to go on increasing the speed.

In the sector of laser disk drives, the little ones are not in fashion - the integration rate is currently going up, so they follow the same route like embedding read slots for various flash cards and increasingly migrating to the combi storage solutions like DVD-RW/DVD+RW due to the lack of unified established standard. Of the most fresh trends, there are plans to start production of disk drives supporting the DVD+R(W) 8X, with respective disks for them already this autumn. Moreover, 16x DVD-recorders are already looming in the distance. At least, this speed should be provided by a 220 mW 650 nm Sony laser whose mass production is to start next summer.