*scottstuff*

The official Xbox360 prices are out. There will be two models:

• For $299, you can get the basic model with no hard drive, no remote control, and a wired controller.
• For $399, the premium model includes a hard drive ($100), wireless controller ($50), remote control ($30), and HD cable ($30).

Microsoft hasn’t announced a launch date yet, but the rumor is that it’ll be available on Black Friday).

Somehow I missed this earlier this month–Gigabyte has announced a $50-ish PCI card that takes up to 4 DDR DIMMs and acts like a SATA RAMDISK. It has a battery that supposedly lasts 12-16 hours and will recharge via the PCI standby power line.

I’ve seen a bunch of people excited about using this as a boot disk or a Windows paging disk, but personally I’d love to see this used as an external journal for EXT3 filesystems. For some workloads, this would result in huge performance boosts for an amazingly small amount of cash. It’d be nice to have more battery life (36-72 hours would be ideal)–my personal record for a home power outage is 13 hours. All of my work-related outages have been brief, except for the facility in Manhattan that was dark for about a week in September of 2001.

via Ambient Irony

Update: I’ve been thinking about this, and the whole thing would be massively more useful with a couple small additions. First, add a Compact Flash socket to the board, and then update the ASIC that runs the board so it will copy the contents of the RAM onto the CF card after an hour or two without power. Then copy it all back when the power comes back up. You should be able to buy 1 GB of DRAM and 1 GB of CF flash for around $150; adding $100 for the PCI card give you 1 GB of seriously non-volatile memory for $250. I’d probably make them a standard feature in every server that I bought, just for the performance boost.

The Inquirer reports that Intel’s Tukwila chip is going to have an on-board memory controller, just like all of AMD’s newer chips. Tukwila is a multi-core Itanium, and is due sometime in 2007; the Inquirer suggests that Xeons will probably get on-board memory controllers in the same basic timeframe, simply because this will let Intel use the same controller chips for both Xeon and Itanium systems.

Assuming that the rumor is true (and considering how well AMD’s on-board controller works, I’d be surprised if it’s not), Intel will probably end up putting 4-6 FB-DIMM channels per CPU; since each channel’s good for around 10 GB/sec, a dual-chip system could potentially have 120 GB/sec in memory bandwidth. Even better, it’d be possible to build a high-capacity server with 48 DIMM sockets spread over the 12 channels; with 4 GB DIMMs, that’s 192 GB in a relatively simple box.

This assumes that multi-CPU systems remain common; given the way that multiple core systems are progressing, I’m not sure that there will really be a market for commodity multiple-CPU-chip systems after 2007 or so–if you can get 8 cores on a single chip, why would you pay the complexity cost of adding more chips, except for really high-end stuff? Even today, compare the cost and performance of an Athlon 64 x2 vs a system with 2 single-core Opteron 2xx chips–the Opteron system will have a bit more memory bandwidth, but they’ll have similar performance on a lot of workloads and an Athlon 64 x2 with cheap motherboard will be cheaper then most dual-CPU Opteron motherbards, never mind the CPUs.

Dual-CPU systems have been the bread and butter of the PC server world for the last 5-7 years, but I doubt that they have more then another two years to go before they fade into the sunset. Personally, I’d much rather manage a handful of single-chip 8-core clustered, virtualized (where virtual environments can migrate between physical systems under explicit admin control) systems then a smaller number of 2-4 CPU 16-32 core systems.

The Inquirer has a photo of one of Broadcom’s upcoming SAS (Serial Attached SCSI, basically SCSI over the SATA physical layer) cards. The interesting thing about this card is that it’s both a PCI-E and a PCI-X card–you can flip it over and plug it in either way.

This isn’t the first time I’ve seen this sort of thing done–there were ISA/MCA cards on the market for a while in the late 80s–but they’ve always been extremely rare. I doubt Broadcom will have this card on the market for very long, because it’s almost certainly cheaper to make two different models then one model with two different interfaces.

I’m really interested in seeing how SAS hardware is priced, because it could be extremely useful in low-end servers. Unlike old parallel SCSI, SAS is a point-to-point network–no daisy-chaining drives on a single cable. Unlike SATA, though, SAS is designed to support “fan-out” devices, so you can plug multiple drives into a single controller channel. Supposedly it’s possible to plug SATA drives into SAS controller; if it’s possible to plug SATA drives into a SAS fan-out enclosure, then we’d get the best of both worlds–the ability to buy big, cheap (but slow) SATA drives and the ability to hang a dozen or so drives off of a single server without needing a dozen different cables. I don’t know if any vendors will be aggressive enough with their pricing to make this cost-effective, though.

The parts for my latest home PC arrived yesterday afternoon and Gabe and I spent a couple hours assembling them into a workable system. It’s amazing how much help a 4-year-old can provide, even around delicate PC parts. I now have a working Athlon 64 3000+ (S939) system with 1 GB of ram and a reasonably large amount of disk space sitting on my desk. I’m going to be using this for Xen testing, but I’ll write more about that later. For now, I want to concentrate on the hardware.

I went out on a limb a bit when ordering this system, because the motherboard I picked, MSI’s RS480M2-IL, is the first motherboard on the market with ATI’s first Athlon 64 chipset (the Radeon Xpress 200), and Google doesn’t give any clear Linux success stories for the motherboard or chipset. However, this is the only socket 939 board that I could find with on-board video, and I really like on-board video for servers. It was also quite a bit cheaper then buying a comparable board plus an AGP video card. A bit of poking around suggested that the SATA ports might be trouble, and it was unclear how well X supports the on-board video, but I don’t really care about either of those for this system. The parallel IDE ports and Ethernet are the only really important parts for me.

So, after installing all of the hardware, I burned a new Ubuntu install CD and gave it a spin. It booted up okay, found the network, found the IDE hard drives, and installed without any serious problems. Ubuntu’s install CD doesn’t seem to have a driver for ATI’s IDE chipset, so I was stuck in slow PIO mode, but it still worked. Once the install finished, I rebooted and watched Ubuntu try to add all of Gnome and OpenOffice to my nice little server system–yikes. After stopping that, I installed gcc, downloaded the source for Linux 2.6.11.4, and build a new kernel.

After booting the new kernel, almost everything looks okay. Here are the drivers needed for this hardware:

• IDE: ATI IXP (in stock 2.6.11)
• SATA: libata’s sata_sil driver detects 4 SATA ports. I have no SATA drives to use for testing, though.
• Ethernet: 8193too (in stock 2.6.11)
• IEEE1394/firewire: OHCI1394 (in stock 2.6.11). Only lightly tested, but able to mount disks connected to FW DVD burner.
• USB: EHCI (8 ports USB 2.0)/OHCI (4 ports USB 1.1). Looks okay, but untested.

I’m currently fighting two problems:

1. Massive clock skew–the system clock is running twice as fast as it should. This is usually a power-management issue or a BIOS bug. A lot of new systems suffer from this, and it shouldn’t take too long to fix.
2. The system won’t reboot cleanly. Linux shuts down okay, but the system hangs and I need to hit ‘reset’ before it’ll reboot. This is probably related to problem #1.

Update (3/18/2005): Disabling the APIC fixed the clock problem, but not the reboot problem. I tried changing a number of power management settings without success. Most likely, the APIC will start working with a future BIOS revision. This problem seems to be preventing me from booting Xen right now, but that’ll probably be fixed by a new version of Xen in the fairly short term.

No matter how much you like Fry’s, 4 visits in a single week is just too much.

I think I’ve ODed on Fry’s. At this point, all I see when I walk in the door is a Walmart with their core demographic switched from rednecks to rabid gamer wannabes. When looking for a cheap, non-tacky PC case on Saturday, I asked a wandering sales guy about cases without plexiglas windows in the side and he replied “we don’t have a lot of demand for those. Pretty much everyone wants to show off their mods.”

Hmm. Looks like both modfree.com and modfree.org are free. I see a merchandizing opportunity here. If anyone wants to pick them up and run with them, let me know–I’ll be first in line to order the “my PC is beige” t-shirt.

The Inquirer is reporting that the CompactFlash trade group released a newer CF spec at CES, bumping the top speed for CF cards from 16 MBps to 66 MBps. This should be great news for photographers–most DSLRs use CF cards, but they’ve been falling behind in the flash speed race. For example, on Canon’s newest 1D-series cameras, the camera’s SD slot runs rings around the CF slot. In fact, the SD slot on Canon’s new cameras seems to be faster then any CF slot on any camera, so it’s not just an issue for Canon’s CF implementation. CF has been falling behind; hopefully this speed boost will let the next generation of cards and devices double or triple their CF transfer speeds.

It’s not widely appreciated just how many different modes of operation modern CF cards have. They’re basically miniature PCMCIA cards, with their own ISA-style IDE controller built in. They’re also IDE devices–you can get an adapter to connect them directly to your motherboard’s IDE interfaces. In addition, modern CF+ cards have a USB interface onboard. I think there are a couple other modes of operation as well, like legacy PCMCIA flash stuff, but I’m a bit hazy on the details.

In spite of all of the complexity, they’re still the cheapest type of flash media on the market.

A friend just sent me a link to this week’s ads for Fry’s Electronics in Seattle. This has been a long-running irritation for both of us–unless we go out and buy a paper, we don’t know what Fry’s is selling this week. It seems ridiculous that they don’t post their sales online themselves; instead, this link above comes from one of the two major Seattle papers; they’re both hosted in nwsource.com.

While it’s nice to see their ads show up online, it’d be even nicer if they were actually usable–as it is, you get an unreadably small image of the ad; you can click to zoom, but the zoomed image only shows a single segment of the ad, usually one or two items. So, to read the whole ad, you’d have to load at least 100 distinct images. Nice work, guys.

CNet has a summary of AMD’s latest analyst meeting, including details on future chips:

On the high end, AMD will release chips with two processing cores in 2005 and then follow in 2006 with chips based around a new chip core code-named Pacifica.

The company is relatively tight-lipped about Pacifica, but said it will be a dual-core chip that also contains virtualization technology–which allows a computer to run multiple OSes–and a security technology called Presidio. Pacifica will appear in desktops, notebooks and servers in 2006. AMD says it will also come out with a new ultra low-power chip for notebooks.

The article also mentions that Intel has been talking about virtualization lately. Considering that PCs have been fighting with virtualization since the 386 was first introduced, I’m amazed that it’s taken 20 years to get full-blown virtualization support in mainstream PC chips. Anyone who’s been around PCs for a while remembers the whole mess with 386–they could run multiple DOS programs at the same time, given a decent vm86 environment, like Desqview or Windows/386–but as soon as you tried to run something that needed 286- or 386-specific features, the whole house of cards came tumbling down, because the 386 couldn’t virtualize itself. Neither could the 486, Pentium, or any of the other x86 chips that have shown up since then.

Of course, we’re better at cheating now: programs like VMWare and Xen have shown that it isn’t really that hard to work around the CPU’s virtualization problems, but you end up paying a price. With VMWare, it’s performance; with Xen it’s patching the guest OSes to not use specific CPU instructions.

Even once that CPU’s been virtualized, the hardest part remains: virtualizing the rest of the machine. Xen’s approach is very open-source centric: they require the guest OS to be ported to Xen, including Xen-specific drivers, rather then emulate specific PC hardware in the virtual machine monitor. Long-term, that’s probably the most reasonable way to handle things, at least in the open-source world.

I’m looking forward to AMD’s new offerings. Pity we have to wait more then a year for them.

According to CNET, Intel has dropped their plans for a 4 GHz Pentium 4 chip, replacing it with a 3.8 GHz chip with a 2 MB cache. Intel is spinning this as a deliberate move to distance themselves from the “more MHz is better” mindset:

Behind the shift is Intel President Paul Otellini, who wants the company to move away from focusing on increases in chip speed, measured in megahertz, as the primary way to increase performance. Intel has talked about such a shift for years, but remained fond of the clock-speed approach until recently. Speeches by executives about moving away from megahertz were often closely followed by announcements of faster chips.

Of course, the spin is wearing a bit thin on this–if Intel could release a 4 GHz P4, then they’d jump at the opportunity. It’s certainly cheaper to produce P4s with 1 MB of cache then with 2 MB; replacing their entire 1 MB line with 2 MB models will lower their profit (assuming that the replacement chips sell for more or less the same price).

On a similar note, Om Malik points out that Intel’s latest quarterly earnings were quite a bit worse then Intel’s been spinning.