by Andy Yaschenko
05/11/2004 | 11:21 AM
AMD has been enjoying a superb vernal season – all three months of it. Well, that’s normal: when Intel is trudging from one tech process to another, AMD is flourishing and looks winning against the competitor. After Intel has polished its process off, the companies exchange their roles. This is an old and natural cycle, like the sequence of yearly seasons.
<%BANNER[article]%>Today we have an AMD season on the calendar, even despite the fact that the company launched no new processors in March, but just offered new revisions of old samples: a 2.4GHz Athlon 64 3400+ with 512KB of L2 cache, a 1.8GHz Athlon 64 2800+ with 512KB of L2 cache and 2.0GHz Athlon 64 3200+ with 1MB of L2 cache. Yes, AMD now has the opportunity of releasing models with a reduced cache size, leaving itself headroom for the future. What if we enlarge the L2 cache to 1MB in the 2.4GHz Athlon 64 3400+? Right, AMD will get an Athlon 64 3600+ into its assortment. And after this model they should master the 90nm process at last.
So far AMD has been busy preparing for the inevitable and imminent transition of the Athlon 64 to Socket 939, scheduled for May 25 when 3500+ and 3800+ models are to be announced. In the first case, the one hundred points above the 3400+ is only due to the integrated memory controller – it is dual-channel now rather than single-channel.
There will of course be a 3700+ model (minus the same hundred points due to the single-channel controller) for Socket 754, but this platform is steadily sinking into the low-end sector. There are rumors again about the upcoming realization of the Paris core: the Athlon 64 with 256KB of L2 cache and the 64-bit addressing capability locked will come under the old brand, as an Athlon XP. It will be compatible with mainboards on new chipsets, though, - unification rules!
AMD is also pushing forward its most successful product, the Opteron. Particularly, the economical 35W model looks very appealing: Angstrom Microsystems has already unveiled four-way 1U servers, HP and Sun intending to offer 1U products on such chips soon.
Opterons with the x50 index are expected soon, solely due to higher clock rates (up to 2.4GHz), but AMD’s more ambitious plans are waiting for the next year: dual-core Opterons in the same Socket 940 package. So, 90nm is enough to pack two chips connected via a HyperTransport bus into the same volume? Or do they mean 65nm?
Anyway, the Opteron is such a success (also because Intel doesn’t offer a worthy alternative) that even the unbreakable fortress of Dell succumbed and expressed a desire to produce servers on this processor. Well, AMD was inside before that – it turned out in April that Dell had been secretly shipping Athlon-based computers to its preferred clients.
AMD is also building up its production muscles. In the middle of the month they reported about a celebration scheduled for the middle of May: the first phase of building Fab36 is over, its walls are ready. Then they will start installing equipment compliant with the third version of the APM technology pack used by AMD (Automated Precision Manufacturing). By the way, the company has established APM centers at Fab25 and Fab30 to accelerate its development. Also in April, AMD said it had created a new division, an Indian engineering center.
But let’s get back to the fabs. One of them will have started its operation by the end of this year. This is only a fab for packaging and testing chips, though, which AMD is going to build in the Shanghai region, next to Intel’s analogous fab. As you see, everything – even packaging and testing – moves from Philippines and Malaysia to China nowadays.
Fab30 reported a success, too. In April, it started test production of 90nm Opterons! That is, the company is checking out how well the equipment handles production of 90nm processors, rather than produces a few samples. They promise to start shipments of the product in the third quarter, but this seems to be an overoptimistic claim, considering the sad experience of Intel in this area.
Now, let’s throw some bad news into the heap. Intergraph didn’t stop at defeating Intel in court, but tried to nibble at AMD, successfully. The company got $10 million on the spot and will have 2% of AMD’s profits in the next three years, but not more than $5 million annually. Well, AMD will hardly earn so much money, considering that it got only $45 million in the last quarter. This is bad for Intergraph because AMD itself should be happy with this result, especially with the sales amounting to $1.236 billion. The processor maker only enjoyed such sums when the first Athlon was at its evolution peak!
This fact is confirmed by the analysis from Mercury Research for the x86 CPU market: AMD’s share grew from 14.7% to 14.9% in Q1, while Intel’s dropped from 83.7% to 83.6%. This is not a good tendency for Intel, especially as the launch of the Prescott happened back in February. On the other hand, AMD’s main blow occurred in the field of server processors, rather than in the desktop market. In the server sector, the company grew from 3.1% to 4.8% and is going to increase its share further in Q2.
Intel’s performance in the last quarter was not very bright. It was not downright bad, too, – with $8.1 billion sales and $1.7 billion profits – but these numbers are 7% and 20% lower than in the previous quarter, respectively. On the other hand, this looks more like a seasonal slump. The results for the current quarter are expected on the same level – Q2 is traditionally a dead season.
In the meantime, Intel uses the temporary lull for reequipping yet another fab, the Arizona Fab12, for 300mm, aiming at 65nm already. This will cost Intel a couple of billion US dollars – not much for a giant who can earn such money in a quarter.
Time is also passing by in brawls with various controlling organizations. In April, trade commissions of Japan and the European Union, as if in a conspiracy, started investigating Intel’s allegedly non-competitive ways of doing business. Intel is claimed to make the customers buy their processors without an alternative. Two identical accusations in two parts of the globe may be a coincidence, but may reflect the real Intel’s methods?
So what about those non-alternative products they make us buy? In April, the 3.2GHz Prescott left paper to be realized and offered. So we don’t see only the 3.4GHz chip among the list of products announced back on February 2. As a kind of compensation, the Japanese retail net reported availability of the 0.13-micron 3.4GHz Pentium 4 Extreme Edition for over one thousand bucks. Curiously, this chip consists of 178 million transistors, while graphics cards on the much more complex GeForce 6800 Ultra chip (about one third more transistors) cost twice less…
By the way, the 3.2GHz Prescott in Japan costs only $310, although it is still tight if we take the GeForce 6800 Ultra as a reference point. In June, the 3.6GHz Pentium 4 560 is expected, officially priced something like $640. There’s only one consolation – it will be accompanied by the 2.8GHz Celeron priced the reasonable sum of $117.
Now what about Intel’s own 64-bit processors? In April, there appeared new 1.4 and 1.6GHz versions of the Itanium 2 with 3MB of L3 cache. Curiously, the performance grew by a quarter over the 1.4GHz Itanium 2 with 1.5MB of L2 cache, while the manufacturing cost, according to Intel, dropped by a quarter, too. Intel’s plans about the merger of the Itanium and Xeon lines are interesting too – the company is going to have a unified platform for these processors by 2007 and the Itanium is expected to surpass the Xeon by the price/performance ratio by that time. Intel’s employees do speak about an inevitable convergence of prices for the two processors in the future.
By the way, these CPUs have got rid of one serious competitor in April. Sun stopped developing UltraSPARC V and Gemini processors to focus on some mysterious high-performing computer systems of the next generation. That’s sad and I only hope this is not an indication of a nearing downfall of the renowned player in the processor market. After all, the money Sun sued from Microsoft is enough for a serious restructuring. The company seems to have started restructuring already, and we may yet see it rebirthing like a phoenix.
Transmeta has created itself another image: of an Alice falling down a bottomless tunnel. The company ended another quarter with a usual loss of $23.4 million with sales amounting to $5.2 million. They have about $113 million on their bank account, which will last the company for a year more of such business. On the other hand, Transmeta has luckily chosen a technological partner, Fujitsu, who started shipping samples of the 90nm Efficeon, and has licensed their energy-saving technology to NEC which seems to be willing to use it in its chips. Hewlett-Packard also announced end of April that it would produce 1U servers on the basis of the Efficeon. They are gonna make it perhaps?
We are still waiting for the new generation of Intel’s chipsets, i915/i925. Originally expected at the end of March, they first moved to the end of May, then to the fourth week of June. If we put aside versions about problems with the 90nm tech process and unavailability of the LGA755 Prescott, the delay may be caused by some troubles with chipsets and mainboards proper: with the socket, DDR2 performance or stability of external PCI Express graphics cards. Something of that may really be true, but this is no good anyway.
Extreme people only have the LGA755 socket so far: the notorious Soltek SL-865Pro-775 appeared in April in the Japanese retail net, priced at $155. This is not much even for an ordinary i865-based mainboard coming with a full accessories set. Once again, all other innovations, save for the LGA755 socket, won’t be really needed in this year.
For example, ATI is going to unveil its DDR2 chipsets in the next year only, reasoning that this memory won’t be necessary in this year because of its steep price. The next new chipset from the company (not counting the RS350 in, which is a revision of the RS300), the RS400, will appear in the middle of 2005, with DDR2-667 and PCI Express x16 support. This looks reasonable, especially with respect to DDR2. The PCI Express bus may become necessary a little bit sooner, so ATI will unveil the RS480 and the RX480, Athlon 64 chipsets with support of this interface, in this or next quarter.
Overall, it looks like AMD hit the aim by integrating the memory controller into the Athlon 64. At least, it’s now clear that there won’t be a quick transition to DDR2, so the basic disadvantage of the controller is negated (it’s difficult to implement support of new memory types), while all advantages (performance!) remain. They will be doing this DDR2 transition without any haste throughout the next half a year.
VIA Technologies is of course faster than ATI – they started sampling their PT890 and K8T890 chipsets in April and are going to start mass shipments of the first of them in June. So it is quite probable that you’ll see PT890-based mainboards in shops sooner than i915-based ones. As for platforms for AMD processors, new faces showed up here. It is rumored that Sun and ServerWorks are intending to offer chipsets for the Opteron. Well, considering the recent advances of AMD in this field, this news seems quite credible.
These are all plans and perspectives, though. As for real products for AMD processors, there’s only the nForce3 250Gb – MSI, EPoX, Chaintech and ASUS all announced mainboards on this chipset in April.
There was scanty info about other chipsets. Well, it’s hard to regard seriously integrated solutions for the Athlon 64 like the Gigabyte GA-K8S760M mainboard on the SiS760 chipset, selling for $120-130 in Japan or the K8M800-based DFI K8M800-MLV mainboard, announced in late April.
DFI behaved overall weirdly in April, introducing such “novelties” as PT800-AL, P4X533-ALE and 661FX-MLV on chipsets that you know of. Well, Acorp was no better than that with its 4PMMN (P4M266A) and 4848P (i848P). Soltek and MSI also took part in this retro-show, being the first to offer mainboards on the dual-channel, but not very necessary nowadays KT880 chipset.
That’s not all as we have a whole weird platform, VIA’s EPIA. With all its ambiguity, it is developing, although not through its chipsets. In April, the speed of the C3 processor grew by 200MHz to 1.2GHz and there appeared a version of the mainboard without any lead. How do you treat this against the background of “serious” mainboards on any of the above-mentioned chipsets?
Of course, this is a purely niche solution, but anyway niche solutions differ. There’s absolute exotics like Transmeta’s Crusoe and there are quite operational workhorses like the server Canterwood-ES (E7210) and Lindenhurst (E7520), coming to the market on MSI’s MasterX-FS and MasterX-FA6R, and Supermicro X6DH8-G2 mainboards that MSI showcased in April in Japan.
Well, it’s time to wind up this section now that we’ve got to niche solutions. A bit of statistics for March: ASUS (with ASRock) with 3.6 million shipped boards is on top, followed by ECS (1.41 million) and Gigabyte (1.40 million). MSI comes next with 1.25 products shipped. It will be interesting to watch the fate of the second place in the next months – will Gigabyte, which has been dwelling on the fourth position, get entrenched there?
It was a crazy month in this market as concerns its main topic – memory prices. The March growth transformed into an ordinary exchange agitation, with prices flying high only to fall down soon. Something like that happened in April. At first, the price jumped from $5.6 for a 256Mb DDR400 chip to $6.66 in the middle of the month and then immediately rushed down, to $5.35 at the start of May. This wide amplitude was an opportunity for some players to earn some money, including memory makers themselves.
Now the situation should be returning into the groove as even those manufacturers that had technological problems are solving them now. I mean the Nanya-Infineon alliance in the first hand, as their troubles with installation of 0.11-micron equipment may have contributed to the recent crisis. They planned earlier to put their joint venture Inotera into operation in Q1, but now this fab will only sample first wafers in this month and will have only started its normal operation by the end of the year. Nanya’s upgrade of its own two 200mm fabs to the 0.11-micron tech process is also going on with delays. As a result, Nanya is simply incapable of meeting the demand from its major OEM clients. Anyway, the company is going to produce up to 80% of all its DRAM chips with the 0.11-micron process in the end of 2004 (against today’s 10%!).
Infineon, on the contrary, said it would reduce DRAM production in two, from 50 to 27 million chips, switching to other memory types. This announcement also contributed to the price growth. But after the growth peak was over and the prices plunged lower than they had been at the beginning of the month, the company imperturbably reported good quarter results, saying that it had no problems with the 0.11-micron process whatsoever, that it wouldn’t reduce DDR production in May in spite of rumors and, moreover, that it would finish building its 300mm fab in the USA.
Micron seems to suffer most from transitioning to the new tech process – the company was surely the worst of all in the first quarter. Like Nanya, it found itself unable to satiate its clients who started moving their orders to Micron’s competitors.
Only the Koreans help keep the prices low. Samsung has successfully switched to 0.10 micron, solving the problems with the dielectric and increasing the chip yield to an acceptable level. Hynix finished the first quarter with a profit of $253 million and said it would change its policy from May, paying more attention to the spot market. It means that the prices there will continue falling down, if nothing extraordinary happens.
Now that we have surveyed the market situation, let’s move on to end products, to memory modules. DDR2 is being announced as if the manufacturers didn’t care a bit about where we might use them (well, that’s really not their problem). Anyway, Crucial, TwinMOS and Adtec all joined the DDR2 camp in April. Interestingly, many companies are offering PC2-5300 modules on DDR667 chips, not yet ratified officially by JEDEC.
Veterans are securing their positions in the market: 512MB modules from Samsung (PC2-3200, $330) and Micron (PC2-4300, $420) were spotted in the Japan retail net. They also announced one-tier registered PC2-5300 2GB modules on 512Mb chips, which came instead of earlier-announced modules with chips places in two layers.
Overall, the arrival of DDR is not a very instantaneous event (you can see it from the prices), and customers shouldn’t be very sad about that. After all, DDR SDRAM handles all the jobs well enough and even continues developing. In April, OCZ again showed up in this field enhancing its Enhanced Bandwidth initiative, which is not a marketing trick, but a set of real technologies that allow increasing the real performance of memory by reducing data access latencies, although keeping the same frequency.
They also released EB modules on DDR466 chips with a peak bandwidth of 3.7GB/s, so 500MHz version of this product should come here soon. In parallel, the traditional EL series is evolving: this memory is made from carefully selected chips and features low latencies. In April, there appeared 550MHz PC-4400 modules of such memory.
The announcement of the GeForce 6800 was undoubtedly the most important event of April. It seems like NVIDIA came up with a practically blameless product for the first time since the TNT, a product that can be truly called a new-generation one: the NV40 provides nearly twice the performance of the NV35! Of course, it comes at a price: two hundred million transistors (central processors are left far behind in this regard!) and an energy consumption of over 70W. Anyway, having spent over $400 million, NVIDIA has fully rehabilitated itself after the last-year fiasco with the NV30.
Graphics card makers reacted with an unsuppressed enthusiasm (well, they would do it in any case, with any new generation of chips, like they did with the NV30), all announcing their products on the new GPU. Even Gigabyte who has made friends with NVIDIA just in time for the release did so. Anyway, GeForce 6800-based cards are expected in retail no sooner than June, so ATI has chances of outrunning NVIDIA in the race for the shops.
As usual, a professional analog of the consumer chip was also announced, the Quadro FX 4000 (curiously, it keeps the FX suffix). Evidently, with the performance growth, the Quadro, once only a nominally professional solution, starts looking more like a pro product. It is especially so as the traditional players in the field, like Matrox, Number Nine and 3DLabs, are passive.
Being the leader of in the field is good and all, but NVIDIA’s share of the market is mostly due to its presence in the mainstream sector, like the GeForce FX 5700 and thereabouts. In April, some activity was felt in this field where card manufacturers caught at NVIDIA’s recent offers: the GeForce FX 5500 and, especially, the GeForce FX 5700 Ultra with support of DDR3 memory. Such cards were announced in March and appeared in stores in April, with a nice price/performance ratio. The GeForce FX 5700 LE, like an in-between product, also enjoys popularity among the manufacturers.
As a result of successful moves in this sector, NVIDIA starts regaining what has been lost to ATI Technologies in the recent months. The Canadians don’t seem to be ready to give up, though. ATI’s ambition feeds on its recent successes and the last quarter should add confidence to ATI, too: $463.3 million sales and $47.6 million net gain.
On the other hand, the event that would break the balance to ATI’s advantage doesn’t happen – MSI still can’t start producing graphics cards on ATI’s chips. In April, we saw photographs of the cards, but there was still no official announcement. Maybe they are waiting for Computex? By the way, the second thing that may happen at Computex is an announcement of the RADEON 9550 for the world market, not only for the Asian one. This chip may make a good competitor to the GeForce FX 5700 LE – it’s silly to leave it for Asia only.
I don’t say anything about the RADEON X800 here as we had only rumors about it in April – let’s better wait for the May results. As for other rumors about hi-end products from ATI, some manufacturers are said to be readying RADEON 9800XT-based cards with 128MB rather than 256MB of memory onboard. This will affect their price, but shouldn’t tell much on their performance.
That’s not all. In April we heard some news from…Matrox! It announced three new products, but all three are multi-monitor solutions for specific market niches like systems for watching satellite snapshots. There’s no mentioning new chips and it seems like Matrox didn’t want to get back to the market with the Parhelia either. They just made the graphics core a bit stronger for using on these exactly niche graphics cards of their own manufacture.
The month was all boiling with news as if reacting to the arrival of the new generation of graphics chips. First of all, we saw numerous new products that prove the rising tendency: matrixes are now improved through reduction of the pixel response time, rather than through better brightness, contrast ratio and so on.
For example, 12msec responsiveness is quite ordinary for the basic 17” diagonal. Cheap models like the Prestigio P175 may have 16msec at worst, although this number was considered a record just a little while ago. So such products as e-yama 17JN1-S or Shuttle XP17 with their 25msec produce a feeling of slight bewilderment. Our choice is something like a BenQ FP783 (although the integrated web-cam may be a superfluous feature), NEC 179M or S, or Samsung SyncMaster 710T – all with 12msec responsiveness.
Times are slower with 19” monitors. Well, 12msec are there, for example in the NEC RTD194S, but it is hardly a norm because there are many models which focus on brightness and contrast ratio, rather than pixel speed. For example, take the Samsung SyncMaster 910T with 25msec and 800:1 contrast ratio (against 400:1 in the RDT194S).
Longer diagonals still don’t have any common norms. The parameters of different models may differ greatly – all manufacturers do as they think right. For example, consider the 21.3” ProLite H540S from Iiyama with 25msec response time, 500:1 contrast ratio, 280 candela/sq. m brightness, or the 20.1” LG L2010B with 16msec response time, 400:1 contrast ratio and 250 candela/sq. m brightness. It’s hard to tell which parameter may be viewed as a norm.
But I can definitely say that the prices will grow in this market, and not only for LCD monitors, but also for CRT devices. In April, the manufacturers increased them by 2-3% and are going to do a 5-7% addition in May due to the deficit of key components. And this deficit may linger throughout the second half of the year. The same goes for LCD monitors: prices for color filters for the panels are growing and thus panels are also growing in price. One of the major manufacturers of panels, Hannstar Display, promises a price growth of over 5% for the second half of the year.
This doesn’t sound very optimistic, yeah? We may turn to OLED displays for better news. Epson showcased a prototype of 12.5” OLED monitor at the EDEX’2004 expo in Tokyo. There’s only one problem with this device – it takes this monitor only 1000 hours for the brightness to degenerate in two times. Anyway, we’ll probably hear first announcements of PDAs or even sub-notebooks with OLED screens closer to Christmas.
Although the market of classical drives has been a place of stagnation, there’s unusually much news here. Maxtor revealed external QuickView HDDs, claimed to be optimized for household electronic appliances. This was the only announcement in April, while two previously-announced drives made it into the market: the 400GB Hitachi Deskstar 7K400 ($480) and the 200GB Seagate Barracuda 7200.7 SATA ($170, with Native Command Queuing, which makes this model much more interesting than the leviathan from Hitachi).
Other, full-size news articles don’t have anything to do with real products: the ratification of the 3Gb/s SerialATA II standard and Seagate’s promises to introduce hard disk drives with a kind of hybrid interface called FATA (SerialATA II + Fibre Channel adapter).
It has now become a tradition that there’s more activity going on in the sector of miniature devices. There are end-products like external 1.8” 40GB drives from Transcend with the USB 2.0 interface or 2.5” internal 100GB devices from Toshiba. There are perspectives for a high competition here: Seagate and Maxtor are preparing to enter the market of small (like 1” or thereabouts) HDDs to join Toshiba, Cornice and Hitachi.
Theoretically, they may meet competition from an unexpected direction: miniature optical discs. Yes, DataPlay didn’t justify our hopes, never really taking off in the market in spite of its obvious advantages (including modest pricing), but LG announced its intention to offer a similar miniature integrated drive for optical discs.
The same LG launched a more ordinary product in April, although it still belongs to novelties – a 12x DVD-burner with support of dual-layer DVD+R discs. If it had a SerialATA interface, like the announced Plextor PX-712SA, and it would be an ideal representative of the last generation of DVD drives. Well, there were no ideal devices among those we saw announced in April. Something was always missing: either 12x burn speed or support of dual-layer discs, or of SerialATA, or both. But if you are no perfectionist, you may get along with a Lite-On SOHW-832S, which understands dual-layer media and costs €160.
Regrettably, Lite-On was only getting ready for the launch of 12x burners in April and only for single-layer discs. After that, after a short pause, 16x burners will come out. From Lite-On, as well as from its main competitor, BenQ, whose DW1600 with support of dual-layer discs is promised for the beginning of June! The manufactures of media are going to speed up production of the discs, too.
As for corporative news, Samsung said it would take to the market of DVD-burners, which the company had neglected somehow. Their joint venture with Toshiba, Suwon, for making optical drives will help the Koreans to start up.
Once in the market, Samsung will have to decide in favor of one of the next-generation standards for optical drives and media. Like a majority of other companies, the Koreans will probably put their stake on Blu-Ray: HD-DVD keeps low, while Blu-Ray is developing actively.
For example, Sony announced a technology of making Blu-ray discs on a paper platter, thus solving the problem of recycling. Pioneer started shipping systems for industrial printing of read-only Blu-ray discs. The apathetic DVD-Forum will surely give this field to the pushing manufacturers, too.
I would usually wind up the review at that, but there was a cause for a P.S. in April. It seemed like the times of drives with removable magnetic media were long gone, but there appeared two such models in April, although based around new technologies! The IO-Data iVDR-20 and Iomega REV 35GB/90GB are in fact twin products, realizing the same idea: take a hard disk drive of one platter, put it into a special case and equip it with a contact pad for using as a cartridge with a special adapter. Unfortunately, the announced starting prices are too high to compete with DVD-burners, for example. We’ll watch for these devices in the future, though.
Of course, you can’t hope for any global events in this field until the launch of the Dothan, but there are small things, too. For example, it’s always good to hear sounds of the manufacturers’ felling prices down. In April, Intel lowered the prices for three mobile processors. The 1.3GHz Celeron M got the most, stepping down from $134 to $107. Simultaneously, at the Japan IDF, they announced new mobile chips, four of them: 900MHz – 1.3GHz LV and ULV versions of the Celeron M and the Pentium M and the full-fledged 1.4GHz Celeron M priced at… yes, $134!
Another good news from Intel, or rather from the Taiwan-based Toppoly Optoelectronics (Compal’s daughter firm, Intel had invested into), was about a new LCD panel that consumes only about 3W of power against 10W of today’s ordinary screens. This is a great achievement for notebooks where each watt directly affects the battery life time.
That’s the more important as this market is developing by leaps and bounds. For example, Hewlett-Packard sold 3.8 million notebooks in 2002, 6.5 million in 2003 and hopes to sell as many as 11 million pieces in 2004! In other words, the company doubles its sales in this area with every passing year! The competitors don’t sleep, either. IBM will abandon the traditional black color scheme for their notebooks, while Toshiba decided to shut down its Philippines fab for notebook production and move its notebook activities from Japan to China. This will negatively affect its sales this year, but should positively influence the manufacturing cost in the long run, i.e. the profit from each sold device.
Again, this market sleeps until the Dothan, although there were quite a few models rolled out in April. Heavy-weight Centrino-based models enjoy success with the manufacturers: they come with a top-end Pentium M (1.5-1.7GHz), with a 15.4” screen, a CD-RW/DVD drive and other parameters, including a weight of 2.7-2.8kg. Specific models are Prestigio Nobile 153, iRU Stilo 3654, and two devices from Acer - TravelMate 6000 and 8000. I guess we can include the new Apple PowerBook model here, with a 17” screen, an integrated 802.11g controller and a DVD-R recorder.
Apple also introduced a notebook in another mass niche: the iBook sub-notebook with a 12” screen. The Wintel camp responded with a volley of Fujitsu LOOX T50H/70H, Sharp Mebius PC-MC30F and JVC MP-XV841. The Centrino-based models (from Fujitsu and JVC) have a 10” screen, while the Mebius not only has a 12” panel, but also an Athlon XP-M 2000+ processor. On the other hand, the Mebius weighs 2.5kg against 1.5kg of the two models on the ULV Pentium M and Celeron M.
Well, not all people need super-portability and not all people are willing to pay extra money for it. Such users are not forgotten: two cheap models targeted at being a replacement of the desktop PC were launched in April. They are the Dell Latitude 100L and the iRU Intro 4015. The first model features a very sweet price/performance ratio – the minimal configuration with a 14.1” screen, 2.4GHz Celeron and 20GB HDD starts at $899.