Saturday, July 17, 2004

New iPods coming next week.

It has been nearly a year since the iPod has been updated (if one excludes its little brother, the iPod mini), but we'll see the announcement of the line's refresh next week, and on the cover of Newsweek no less.

The new iPod will have increased storage capacities, with the flagship model apparently using Toshiba's 60 GB 1.8" drive. However, more importantly, the new iPod will adopt the interface of the popular iPod mini. While the 3G iPod's interface is good, the iPod mini's interface is even better. It's simple and elegant, and I'm glad to see that the mini's big brother will finally get it.

Also, next week will see the iPod mini going worldwide. With new iPods and worldwide iPod minis, one can only guess how high iPod sales and iTunes Music Store download numbers will be this quarter.

Friday, July 16, 2004

IBM talks about G5 yields

IBM had a very good quarter overall, with nearly $2 billion in profits (which is equivalent to Apple's entire revenue for the quarter). In IBM's Q2 2004 earnings presentation, there was also an interesting tidbit about the G5 970FX. Below is a quote from Mark Loughridge, Senior VP and CFO at IBM:
In our 300 mm facility we are doubling output quarter to quarter, first quarter to second quarter, and we have the potential to redouble again in the 3rd quarter. So this is the original ramp that we had planned for, albeit at a one to two quarter lag. Now I also want to point out that we do hundred and hundreds of different chips in our microelectronics division, and we're meeting customer commitments to all of our customers except for one chip. So that's one chip out of hundreds. I need to reiterate: There's only one, maybe two, companies in the world doing this technology, and it really is truly leading edge, where circuit dimensions are measured in single digit atoms. It's the highest performing processor in the world at these power levels. You've seen the performance of our POWER5 chip, real screamers, and our customers are getting performance levels that put them at the forefront of their industries as well. So as I look to the future, in the second half, we see our performance continuing at this ramp and a lot of improvements and confidence in our second half for this technology business. And it's really based on that that we reiterated that we're confident we'll make money in this unit for the year.
There is a lot of PR speak in that quote, but his statements (again) confirm what we knew all along, that IBM delayed Apple for nearly half a year, because of poor chip yields at 90 nm. Also, partially because of this they'll still losing money in their technology group.

Still, there is a silver lining to this. They expect their technology group to become profitable in 2004. Yields are improving and chip output may soon be four times as high as it was initially. (I'm assuming he's mainly talking about 90 nm chips here, since 130 nm yields were already good and satisfying customer demand.) It's also likely that IBM will be able to keep up with G5 970FX chip demand for Apple's Power Mac and iMac lines for the second half of 2004, and it bodes well for a 3 GHz chip release in early 2005.

Apple's 3 GHz will happen, but just 6 months late.

Thursday, July 15, 2004

8X slot-load laptop DVD burner now available.

Panasonic has released a new version of their laptop specific drive, called the UJ-830 Super MULTI Drive (PDF). It's a slimline slot-load laptop optical drive. The specs are:

12.7 mm height

3X DVD-RAM write
8X DVD-R write
4X DVD-RW write
8X +R write
4X +RW write
24X CD-R write
10X CD-RW write

8X DVD-ROM read
24X CD-ROM read

Apple has a long history of using the Panasonic/Matsushita slot-load DVD-R burners in its laptops, and this one should be no exception. I expect to see this drive in the next PowerBook (likely along with a G5) in early 2005. Hopefully, in addition to DVD+/-R/W, Apple will keep DVD-RAM support in the drive's firmware, since OS X has built-in HFS+ read/write support for DVD-RAM. Such support would allow us to use a DVD-RAM disc as if it were a 4.3 GB floppy disc. Also, the 8X DVD-R support will be very welcome. That's just 8 minutes to burn a DVD, on a laptop.

For more information on the various DVD recordable formats, please check out my DVD Recordable FAQ.

Wednesday, July 14, 2004

It's Official: G5 in new iMac

Yes folks, it's official. The new iMac in September will sport a G5.

Peter Oppenheimer, Apple's CFO, said in the Q3 conference call that the primary holdup for the new iMac is a shortage of G5s.

Now on to the rest of the numbers...

This quarter, Apple had $2.014 billion in revenue, with $61 million profit (which includes a one-time $6 million restructuring charge). That's 16 cents per share, slightly beating the analysts' expectation of 15 cents. Apple sold 860000 iPods and 876000 Macs, which represents a year over year increase of 183% and 19% respectively.

The approximate breakdown of Macs sold is:

243000 eMacs/iMacs (including 60000 iMacs)
173000 PowerMacs (including 13000 Xserves)

240000 iBooks
220000 PowerBooks

This is pretty impressive. As expected, the iPod is doing very well, and I would expect them to do even better in Q4 (assuming Apple can meet demand) since the iPod mini is going worldwide in a couple of weeks, and also because we may get a new iPod release soon. However, in addition to great iPod sales in Q3 there were also substantial improvements in Mac sales, including very nice sales of the Xserves (and more than I expected) despite their problems with G5 970FX chip supplies from IBM.

They also separated out iMacs from the eMacs, and the numbers confirm what many of suspected all along, that the G4 iMac is black sheep of the consumer desktop family in terms of sales. For both our sake and Apple's sake, I hope the G5 iMac in September will wow us again into opening our wallets.

G5 Supercomputers Galore, in 2004

VT: Success or failure?

The Virginia Tech supercomputer experiment generated a lot of excitement when it was announced... and it has generated a lot of controversy too. There's no question that with pizza-fueled student labour and 1100 dual Power Mac desktops (not servers), VT got a great Linpack benchmarketing score for the buck. For a mere US$7 million, they got a machine that scored in the top 3 in the world supercomputer rankings and the first educational cluster to top 10 Tflops/s. And they did it with a new platform (Macs) and with a new OS (OS X.2 Jaguar) (and with high-performance Infiniband interconnects that didn't even had OS X drivers until then). That's a remarkable achievement.

However, it has been plagued by problems. First of all, it doesn't seem that System X (aka Big Mac) has actually been used for anything other than benchmarking and basic testing. There had been rumours of ongoing instability issues, which one wonders might be related to the lack of both ECC memory and real hardware monitoring. And of course, the Power Mac based cluster no longer exists. It has long since been dismantled and sold off as high-priced refurbished equipment at various retailers. Indeed, the cluster had to be removed from the latest top 500 supercomputer list for this reason.

VT Redux

Well, as of just a few days ago, the VT cluster is back. It has already been assembled, this time with dual G5 Xserves. Again they're at 2.0 GHz, and again there are 1100 of them. This cluster solves several key problems, with its support for ECC memory and hardware monitoring. It also offers significant power improvements and of course space savings. It is undergoing testing now, and benchmarking (again) will hopefully start next week, which gives them plenty of time to get on the new supercomputer list coming in a few months. If you're in the neighourhood, they're even going to start offering tours of the facility in a couple of weeks.

COLSA does one (or 466) better

With a healthy dose of cash from the US Army, COLSA is building a G5 supercomputer using 1566 dual 2.0 GHz Xserves, which is 42% larger than the VT cluster. Like the VT cluster, The G5 Xserve's biggest draws were the price performance ratio along with the fact that it's a unix-based system. Some may note that the Linpack benchmark speeds may not be all that great because it's using the built-in Gigabit Ethernet as the backbone, not Infiniband cards. However, it's going to be used for computational fluid dynamics and stuff like that, which they say is less dependent on network speed.

UCLA gets in on the action

For about a million bux, UCLA's Plasma Physics Group is reported to be getting a nice "little" cluster of 256 dual G5 Xserves. Unlike COLSA and VT however, these guys do have some experience with Xserve clusters, since they've worked with NASA to create clusters based on the G4 Xserve. Before anyone asks why anyone would use a G4 Xserve for a cluster in the era of fast Xeons, CFD and other code may be amenable to Altivec acceleration, so despite its lousy FP performance, the G4 often is quite nice for this type of work. Similarly, Genentech seemed to love its G4 clusters for bioinformatics work, precisely because of Altivec. In fact, Apple now has a G5 Xserve based Workgroup Cluster specifically geared to bioinformatics types, along with pre-installed Altivec-accelerated bioinformatics applications.

The University of Maine, too

The University of Maine is also creating a 256 dual G5 Xserve cluster, also with funding ($680000) from the US Army. It seems the US military has just as much money to throw around these days as ever, if not more. Assembly of this system has already begun, and you can follow its progress in their gallery.

Xserve sales unprecedented

This has been a remarkable year for the server group at Apple. The G5 Xserve is hugely popular, being aggressively priced, well-packaged, and apparently quite easy to use. Apple does not publish separate Xserve sales numbers, but it has long been rumoured that the G4 Xserve's unit sales ranged only the few thousand range per quarter. These few G5 supercomputers might just equal those types of sales numbers, and that's not including the large numbers of other Xserves that are being sold to small labs and small businesses. (It's rumoured that large clusters represent the minority of Xserve sales.) Too bad it has taken so long. The G5 Xserve was announced in January, but because of IBM's 90 nm growing pains, it has taken until June/July for the G5 Xserve to ship in volume. This is right at the transition from one quarter to another, which will impact on the G5 sales numbers to be announced today. Still it looks like IBM and Apple are over the hump, and the coming quarter looks very positive for the Xserve. I would not be surprised to see shipments of 5-digit units in this quarter. As for the future, Xserve sales will probably just get better in 2005. That's when OS X 10.4 Tiger comes out, which will allow true 64-bit support (with memory addressing over 4 GB per process). Plus, by then, people will have been able to see functioning G5 supercomputers in action, doing real work.

Oh, and one more thing...

While it's not a Mac supercomputer, it's still a G5 based one. IBM has been commissioned to build a 2282 dual G5 blade server based system for Spain, using 2.2 GHz 970FX chips which don't exist on the Mac side. I wonder how long it will take for 2.2 GHz G5s to make their way into Power Macs and/or Xserves. If IBM can put 2.2 GHz chips into blades, then Apple should have no problem putting them into Xserves.

Tuesday, July 13, 2004

POWER5 is out, and it's FAST

IBM today officially announced the eServer p5, which uses their POWER5 CPU.

Why should Mac users care? Well, the G5 is based off the POWER4 series, and it's widely expected that the next-gen CPUs Apple will use will be based off the POWER5. And judging by the POWER5 benchmarks (PDF), POWER5 is extremely fast:

SPECfp2000
POWER4+ @ 1.7 GHz - 1642
POWER5 @ 1.9 GHz - 2576

Despite only a 12% increase in clock speed, there is a 57% increase in performance for the POWER5. Clock for clock, POWER5 is 40% faster than POWER4+. Not to mention it bests the fastest Itanium 2 (2161) Intel has to offer, by nearly 20%. Very impressive.

Some improvements the POWER5 brings:

SMT: Presumably one of the big factors in the speedup is the addition of simultaneous multithreading. I suspect that a POWER5 derivative (G6?) will retain this feature, at least for chips in high end desktops (aka Power Macs). In a nutshell, SMT allows the CPU to be seen as if it were dual processors. (For those of you used to PC hardware, it's called Hyperthreading by Intel.) If POWER5-lite does indeed have SMT, a dual Power Mac would be seen by OS X (10.4 Tiger?) as quad processors. It wouldn't be as fast as a true quad Power Mac of course, but the potential speed boost would still be substantial for certain (multi-threaded) apps.

IMC: Another benefit of the POWER5 design is an integrated memory controller. (Think AMD Opteron for you PC types.) This could also be included in POWER5-lite, potentially increasing memory bandwidth in future dual Power Macs. An integrated memory controller could also potentially decrease overall power usage for POWER5-derived chips in PowerBooks in the (distant) future.

L2: The POWER5 has a larger L2 cache than POWER4+. It's likely a G6 would also sport a larger L2 cache than 512 KB in the G5.

I do find it curious that IBM does not publish SPECint2000 scores for the POWER5 however. Is it because the integer scores are merely mediocre? If so, will that mediocrity in integer performance be reflected in the POWER5-lite too?

So when will we see the POWER5-lite anyway? Well, who knows... Steve Jobs last year stated he expected 3 GHz by now, but we don't know if that 3 GHz was going to be another G5 or something else. I'd guess that the 3 GHz will still be a G5 (970FX), possibly announced at Macworld January 2005, with the G6 not coming out until much later in 2005.

BTW, OS X 10.4 Tiger will include GCC 3.5 in Xcode 2.0, with support for autovectorization. IBM's uber-fast XL compilers will also be updated, including support for autovectorization, by the time Tiger is out. We won't all have to buy new G5s (or G6s) to get faster performance. :)

Contrary to popular myth, the G5 is not hot.

Everywhere I go on the the Mac sites, there is a declaration by somebody that the G5 is too hot to be used in anything other than Power Macs.

Hogwash. Let me lead you through some specifications:

G4 7455 (180 nm) 1.0 GHz: Typical 15 Watts, Max 22 Watts
G4 7455 (180 nm) 1.33 GHz: Typical 30 Watts, Max 40 Watts
G4 7447A (130 nm) at 1.42 GHz: Typical 21 Watts, Max 30 Watts

Of note, Apple publishes in press materials that the G4 7455 1.33 in the G4 Xserve puts out 45 Watts, so the Motorola/Freescale max number might be at bit low, but at least it's close.

G5 970 (130 nm) 1.8 GHz: Typical 51 W, Max ??? (est. 90-100)
G5 970FX (90 nm) 2.0 GHz: Typical 24.5 W, Max ??? (est. < 50)
G5 970FX (90 nm) 2.5 GHz: Typical 50 W, Max ??? (est. 90-100)

In contrast to Motorola/Freescale, IBM does not publish max power ratings. They only publish typical power ratings, and furthermore it should be noted that these numbers cannot be compared with Freescale's directly. However, these ratings are useful as a general guide. It has been estimated by hardware sites that the G5 970 at 1.8 GHz probably puts out more than 90 Watts, almost twice its typical rating. It is suspected that the 970FX at 2.5 GHz has a similar power output, especially given its identical typical power rating. If one were to double the rating for the 2.0 GHz 970FX one would get a max power of < 50 Watts, which is close to Apple's PR material number of 55 Watts. And measurements of the G5 Xserve have put the power output at less than 50 Watts running benchmarks at full blast.

OTOH, the G5 970FX (90 nm) at 1.4 GHz is rated for a typical power of only 12.3 Watts. We might thus guess the max power is < 25 Watts. Extrapolating, at 1.8 GHz it'd probably be close to about 30 Watts max, if the voltage could be kept constant (see below).

But wait, I've previously said that the 2.0 GHz 970FX is 45-50 Watts. What gives? Well, that's a desktop/server chip, and voltage is not a major issue. A deskop/server has effectively unlimited power. However, IBM has indicated that it views the 970FX an an all-in-one design from laptops to server. We already know that the 1.4 GHz chip runs at a lower voltage, and one might guess that there would be low voltage bins of the 1.8 GHz chips too specifically for laptops. Such low voltage chips would likely be more expensive, but that's OK in this context, since the power advantages outweigh the cost.

To get back to the original point, it seems from the specs that a G4 1.5 is hotter than a G5 1.4, and it's also probable that a G4 1.5 is hotter than a G5 1.5 as well. ie. MHz for MHz, the G5 might just be cooler than the G4, at least when we're talking embedded/laptop-oriented chip bins for the G5. Now who's to say IBM is going to make embedded/laptop chips? Well, Motorola for one, since they've been campaigning hard in developer documentation about their claimed advantages of the G4 over the G5. They wouldn't bother unless the G5 was not a real threat for their embedded bread and butter. Of course, I've also shown you the low voltage 1.4 GHz chip specs above, so clearly IBM is looking at this direction. And read on...

I have linked an interesting article in the title of this weblog. In the article IBM went on record to say that it feels that a 30 Watt chip is most appropriate for laptops, and suggests that this is doable up to 1.9 GHz. ie. I might be underestimating the coolness of the G5. And given that the G4 7447A 1.42 is 30 Watts, and that the PowerBooks run 1.5 GHz G4 7447A chips, it's likely that PowerBooks are already over 30 Watts. If IBM's statement is to be believed, this suggests 2.0 GHz 970FX chips for laptops are possible too, for a CPU power utilization relatively close to the 1.5 GHz G4 PowerBooks.

Now, Apple has already said that it won't release a G5 based PowerBook in 2004. If the chips are so cool then why not? Well, it's likely a number of reasons, not the least of which is chip supply, but one of the main issues here in terms of the CPU is not overall power per se, but power density (since the heat is more concentrated over all smaller die). (There is also the issue of higher power utilization of the G5's system controller, but that's a whole other kettle of fish.) But given the numbers above it's a reasonable guesstimation to think that the 2005 G5 PowerBook will reach 1.8 GHz. From the marketing standpoint, this makes sense too, since the PowerBooks are already at 1.5 GHz. A G4 1.5 --> G5 1.8 jump seems perfect.

BTW, this power density issue is also what necessitated the adoption of liquid cooling for the 2.5 GHz G5 970FX. Straight from the horse's mouth: Apple has stated the chip isn't any hotter than the previous chips, but that power density was higher.

So... There has been a lot of numbers here, but the take home message is that IBM has a very nice chip in its hands with the 90 nm G5 970FX, and at lower GHz speeds and low voltage, it's actually quite cool (in more ways than one). You can count on a G5 PowerBook in early 2005, running at speeds which even in 2004 are considered only the realm of Power Macs.

Monday, July 12, 2004

$999 iMac G5? VPC for G5 delayed again.

Well, I'm still getting used to this blogging thing... By accident, I deleted the first and only comment I've gotten. There was this nice little button and I didn't know what it did. So I clicked it, and poof! The comment was gone. Live and learn I guess.

$999 iMac G5

Sulis's comment was about a $999 iMac G5, and how that is likely not possible with a 17" LCD all-in-one, despite the fact that Fred Anderson has already gone on record to say that the US$999 price point (in general terms) is something Apple is looking at.

Well, I agree with Sulis. I don't think it's likely that Apple will make a 17" AIO G5 machine priced at $999. However, I do think it's plausible with a 15" AIO. In other words, contrary to the general wisdom in internetland, I'd be surprised if Apple scrapped the 15" AIO completely. A stripped down 15" AIO with G5 1.4 GHz would still sell reasonably well I'm guessing, for example to institutional purchasers and of course to precisely those consumers looking for a low budget LCD iMac. I wouldn't buy one myself, but I do think the 1024x768 screen size is more than adequate for this market, especially considering the screen is closer to a 16" screen in CRT terms. Well, maybe $999 is pushing it, but certainly $1099 is a reachable price target for this machine.

Some may bring up the eMac, but unfortunately, LCDs are where it's at these days, even in the consumer space, and even at the sub $1000 price point.

Virtual PC for the G5 delayed again

In other news, Microsoft has announced that Windows XP Service Pack 2 will be delayed until August. Considering that Microsoft has said that a G5 compatible version of Virtual PC won't be released until SP2 is out, the absolute earliest we could see the updated VPC is thus August, although it's likely that it won't be until September or later. That will impact some Power Mac users, but at least for the iMac crowd that may not be a big deal, since iMac G5s won't be out until September anyway.

100 000 000 Songs

Oh and I almost forgot: Apple finally reached its 100 million mark for songs legally downloaded from the iTunes Music Store. Unfortunately, we in Canada so far still can't download anything. The wait continues...

Sunday, July 11, 2004

Spotlight: iMac. Paris. Birthday.

These are interesting times for Mac users. It has been a full year now with IBM's G5 in Power Macs, and the trickle down of G5s to consumer machines is beginning.

Given Steve Jobs' subtle hints at the Worldwide Developers Conference and the odd but revealing statements on the Apple Store website, it's clear that we'll see a G5 based iMac or iMac-replacement on August 31 at Apple Expo, to mark the 20th anniversary of the Mac. Yes folks, the first mass-produced GUI-driven personal computer is 20 years old this year. How quickly they grow up. And to think, it was this very event in Apple history that caused me to switch from Apple to Microsoft-powered PCs. And I stayed there until just a few years ago... When OS X 10.1 came out I immediately bought my first Mac. It was an iBook, and I liked it (and OS X) so much that just one year later I upgraded to a 15" PowerBook Titanium. But I digress...

So what will the new summer 2004 iMac include? Well, Apple has already stated that it is a new line, which obviously sounds like a new form factor, housing the G5 970FX. Given that the Power Macs are now all duals and start at 1.8 GHz, a single PowerTuned 90 nm 1.8 GHz 970FX (with its favourable heat dissipation characteristics) seems like a perfect fit for the new iMac. Thus, it wouldn't be too much of a stretch to think that the new top-of-the-line iMac would be configured something like this:

G5 970FX 1.8 GHz, with 512 KB L2 cache
3:1 multiplier, for a 600 MHz bus
256 MB Single-channel DDR333 memory, with 2-3 memory slots
80 GB 7200 rpm hard drive, upgradable to 160 GB
8X SuperDrive (DVD-R/W, CD-R/W)
nVidia GeForce FX 5600 Ultra 64 MB DDR
Airport Extreme ready
Bluetooth ready
Firewire 400 and USB 2.0 ports
VGA/S-video/composite video output
Analogue audio input and output

And yes, I suspect the new iMac is still going to be an all-in-one, with a built-in LCD screen. I'd love to see a headless iMac (aka Cube-redux, consumerified), but somehow I just don't expect it given Apple's recent history. Apple does always continue to surprise us though.

But most importantly, despite the vastly upgraded CPU and bus speeds, I suspect the price of the new iMac will be LESS than what it is now. The current LCD iMac is a design marvel, but it simply is priced too high for its hardware performance specs. The requirement to buy a built-in LCD sure doesn't help cost-wise either, even if it is a good quality LCD with a digital connection (not analogue VGA). Apple must drop prices to compete, and judging by Fred Anderson's price point comments in the recent past, I believe they will. Furthermore, I believe they can compete favourably, as Apple's aggressive pricing of the dual Power Macs illustrates. The Power Macs may not be the absolute cheapest, but they are a very good value. The current LCD iMacs overall are not, but the introduction of a new G5 line gives Apple the opportunity to change this. The success of the new iMac depends on it, and even more so than cool industrial design.

As for the G5 PowerBook, Apple has already told us it won't be available in 2004. That is unfortunate, but a G5 1.8 GHz PowerBook running OS X 10.4 Tiger in the first half of 2005 would be perfect... along with a 6 GB version of the oh-so-sexy iPod mini. Give us that combo, and I'll whip out my credit card faster than you can say Reality Distortion Field™. :)

Well, it's about time, Eug...

So, after years of seeing every idiot on the web posting his/her own weblog, I thought I'd add just one more idiot to the mix.

It will be an interesting experiment...