Slashdot Mirror
The Transmeta Pushme-Pullyou?
tired.cranky writes: "An article on LinuxDevices.com sez that Transmeta is about to ship a quasi-distro slash embedded development toolkit featuring Linus' new super-efficient cramfs and ramfs filesystems. Apparently, a reasonably normal Linux system can be shoehorned into 8MB of storage, with zlib decompression-on-demand and such. It sounds like it could push a fair few hobbyists and embedded developers in Transmeta's general direction, too... and reads nicely next to a Register piece on Transmeta's leaked server initiative. Does one end of Transmeta know where the other is pointed?"
When you're trying to cram as much computing power into a given space as you can, then you'll find that you're limited by heat dissipation. Thus the performance/Watt ration becomes more important than just performance itself. Running lots of slightly-less-impressive Crusoes will end up giving you more computing bang for your BTU, and hence more computing/inch.
It's a neat angle, but then again being able to play solitaire during your entire flight cross-country was a good angle, and that doesn't seem to have panned out. I hope it works, not because I love transmeta, but because I just like seeing cool technology win.
The enemies of Democracy are
It might not be a big deal now, but how about in 6 months or so when the real cost of electricity is passed on to the consumers? If you started paying 10x more for your electricity than you're paying now, will power savings be an issue?
-jon
Remember Amalek.
> Windows 2000 Server is an entry-level solution for running more reliable and manageable file, print, intranet, communications and infrastructure services. Windows 2000 Advanced Server includes additional functionality to enhance availability and scalability of e-commerce and line-of-business applications.
Really. So, if Windows is the superior operating system, and, by logical extension, Microsoft's code is superior, I have two questions:
1. Why did Microsoft use GPLed code in its FTP.EXE program included with Windows? (Load it into a hex editor and look for Regents)
2. Why is Win2K's TCP/IP stack fingerprint nearly identical to a BSD fingerprint (BSD having been around longer)?
Gee. You know the M$ lackeys are starting to wet their pants when they start inundating Transmeta stories with "W2K is good" posts. Especially when the posts only compare W2K with its almost useless predecessors.
Nice going, guys, keep it up. Waste your time flaming Linux on Slashdot while the rest of the world is busy actually writing real code and not flashy interfaces for a buggy platform. Idiots.
The low power, many microprocessor solution has been propounded many times, and it has routinely failed in the marketplace. Why?
People don't care about low power on servers. Even if the power costs and cooling costs are trippled, only a handful are in a position where optimizing for power, not speed is important.
Theoretical peak, N cheap processors is much more powerful than one expensive processor
Then why are sales of low-power servers skyrocketing, especially in San Francisco, which someone from Berkeley should know about. Hint: read the local business pages sometime. I do up here in Seattle, and it's reported here.
There are high-demand servers, mid-demand servers, and back-up servers. For high-demand, constant traffic servers your arguments make a lot of sense. For mid-demand or failsafe rollover servers, the heat cost and power cost are a higher fraction of the operating expense of the server, which is likely a cheaper box with dual drives but chock full o RAM.
Servers don't come in one size or one shape. Some people serve up lots of static pages, some people have worldwide operations. Others have frequent low-traffic periods and may have a primary hit box with some image servers that can afford to have a little lag time and would get a high return on code-replication as it serves the same file over and over and over and over and over.
Just because you don't like Dietzel doesn't make you right. Think before you post, do some research, a few base google queries, check the local trade mags, before leaping to assumptions.
--- Will in Seattle - What are you doing to fight the War?
It might not be a big deal now, but how about in 6 months or so when the real cost of electricity is passed on to the consumers? If you started paying 10x more for your electricity than you're paying now, will power savings be an issue?
Unlike California, where there's a 10 percent rate cap on power prices, both Oregon and Washington states have no such limits and we're looking at 50 to 150 percent price hikes over the next six months, some of which will be permanent.
So, especially in other portions of the Pacific Northwest, this is very much an issue.
--- Will in Seattle - What are you doing to fight the War?
100cpu x 1w/cpu = 100w for the Crusoe's
Ouch.
Of course 100 cpu's need more ram etc. and all that jazz I suppose, but then what do I know, I work with handhelds/embedded systems not huge servers.
Can you imagine a Beowulf Cluster of those?
sorry, but it might have some practical applications.
--- Will in Seattle - What are you doing to fight the War?
I wouldn't throw out my current servers, but I would certainly favor boxes that were otherwise equal but used less power. Why?
1) Why be wasteful? Just because you can afford to do something doesn't mean you should. I pay about 5% more to use 100% renewable power. (Before the latest crisis, I paid 20% less.)
2) It's cheaper. Even under moderate assumptions, you could save a couple thousand dollars a year for a medium-sized commercial web site. That money would be much better spent on more hardware. Or more beer.
3) It's cooler. Not in the sense of hipness, but in terms of temperature. If the Register's numbers for a dual-Itanium server power consumption are to be believed, a couple of those babies would put out more heat as a hair dryer or a space heater.
Judging by Athalon's heat output, the heat output of typical CPUs scales 1:1 with speed. A lot of server rooms I enter are already running a little warm; imagine what it will be like after a round of upgrades to faster and hotter boxes.
Maybe you can afford to pay for the electricity, but can you afford to pay for a massive upgrade to your air conditioning?
And we're supposed to be impressed? Let me break the news here. QNX did it ages ago when they squeezed their kernel, GUI, web browser, mail client and tcp/ip stack into a SINGLE 1.44 MB floppy. Oh, the power of Closed Source!
Your pizza just the way you ought to have it.
The low power, many microprocessor solution has been propounded many times, and it has routinely failed in the marketplace. Why?
Similarly, non-SMP machines have repeatedly failed when compared to SMP machines, because a message passing machine is much more difficult to program. Yes, if you can rephrase your programs to run on a cluster, then you see the impressive possibility in cost/performance, but it is really difficult to program well. People pay the extra cost for SMP machines simply because they can actually program them for a wider variety of applications.
Most problems which can be translated to a message passing structure have already been migrated to clusters of cheap machines. It is problems like databases, which don't map well to message passing, which is why people buy "servers", and it is these problems which are why people buy E10ks.
Dietzel knows better, he is an intelligent scientist, but he is acting like some corporate PR flack. Such dishonesty I can accept from someone who is ignorant, but he knows better.
Nicholas C Weaver, Winged Rat Consulting
nweaver@cs.berkeley.edu
Test your net with Netalyzr
Um, which government would that be? The government which is headed by TWO ex-oilmen? The government which was bought by large engergy concerns? The government whose "help" to California so far has been to offer to remove pollution caps? Give me a break.
Dubya and Vice President Heart Attack are going to screw California to the wall. Since they won't win in California in 2004, anyway (and since virtually all statewide and national officials from California are Democrats), they don't care.
Maybe what California needs to do is stop passing along tax revenues to the federal government. Then we can see if they're willing to help.
-jon
Remember Amalek.
* People don't care about low power on servers. Even if the power
costs and cooling costs are trippled, only a handful are in a
position where optimizing for power, not speed is important.
At LWE, he talked about the Exodus deal. "People don't care" is false, with Exodus being case-and-point. Exodus is limited, by the state of California, as to how much power they can consume and thus how much they can expand. They are reaching their limits right now, with their current architecture. Thus, it's pretty evident that Exodus does 'care' enough about lowered power consumption, enough to make the move to these low-wattage configurations.
That's a good question. One of the side effects of California's recent electricity problems has been a spike in demand for low-power devices, specifically Transmeta-based servers, but the whole device side is another animal.
My impression is that Intel is having problems recognizing how to deal with the appliance picture and small devices, and that Transmeta has maybe an 18 month lead on them. But they're not the only players in this sphere, so this is going to depend on the following things:
1. How fast Transmeta can ship production quantity chips for this market.
2. How well device manufacturers integrate these in a useable way.
3. How interested the consumer market is in these devices.
4. What pricing strategy Transmeta has for this.
5. How far competitors are willing to go with fake media releases, arm twisting, collusion, and rumors to sink Transmeta.
6. Where G Bush and Bill G and their posses have invested their money in this area - if the regulators and the tech money interests are all after Transmeta, the best solution may not necessarily win, unless it can get its own crew behind it.
--- Will in Seattle - What are you doing to fight the War?
I think Transmeta is in a serious slide. A real product needs to come out and be really successful because it's getting close to being a real flop.
The big question (the register's claims aside) is how many crusoes it takes to replace one Intel whopper.
Several things play in transmeta's favor:
1) server code is fixed. This means that the costs due to code morphing don't need to be paid often. This is akin to java's HotSpot technology -- it needs to run a long time to amortize its startup costs. I expect crusoes to work better in servers than in clients for this reason alone.
2) Because they can effectively hibernate when idle, you only pay for electricity for the cycles you use. If budgets are tight, you could envisage a system with under-specced cooling, and temp monitoring -- this would allow you to run at say 80% of maximum constantly, and at 100% for short periods until the temperature rose too high and the processors were forcibly throttled back. This probably fits many usage scenarios very well.
3) If it takes several crusoes to replace one itanium, those crosoes will also be much more fault tolerant, allowing you to continue at a lower capacity while the faulty one is replaced and rebooted.
The transmeta chips *are* low power. For any given desired processing capacity, the transmeta solution will require less power, space, and cooling than the comparable intel solution.
Of course, holding power fixed, you get more computing horsepower than intel. However, most companies decide on the cpu power first, and then accomodate that with space, cooling, and electricity infrastructure.
...it's named 'CramFS' as opposed to FAT
Let us assume a 1u high rackmount mini-server, which is burning 100W of power [1], and which will have a 2 year lifetime before you call it obsolete and scrap it because the space you are renting would be better served by new machines.
Now let us assume a ridiculous power cost of $.50/kwh (note, current rates are around $.10/kwh, and may rise as high as $.15 when all is said and done. Even if you are being charged for cooling as part of the power budget, 50 cents a kilowatt hour is a ridiculous charge).
This means the server costs $.05/hr to operate, or about $900 in lifetime power costs over the 2 year lifetime, assuming vastly overpriced power.
Now given a miracle transmeta server which burns only 50W (after all, you still have to power the ethernet, disks, memory, etc). EVEN IF it has identical performance, it can only cost $450 more for it to be cost effective. If it is lower performance along with lower power (which it will be), there is no point in purchasing it.
[1] Current dual processor, stacked, HIGH end (2x933 MHz PIII) 1u systems will burn ~200W. Couple that with a still excessive but slightly more reasonable power cost of $.25/kwh and the calucation is roughly the same: a lifetime power cost of $900. Something like the BriQ is only burning 40W, but being sold for the remarkably low form factor (can probably fit 4 in a 1U rackmount), not ops/W.
Nicholas C Weaver, Winged Rat Consulting
nweaver@cs.berkeley.edu
Test your net with Netalyzr
The world has changed - there is now a huge demand for 1U rackmount servers to plug into web farms. Even single websites are using web farms for scalability, of course.
A big issue with web farms is data centre space and cooling - Transmeta can greatly reduce this if it fits many CPUs in a single box within the same heat/power budget. Even if they just have one or two CPUs per box, they can make an impact on this.
Transmeta is about to ship a quasi-distro slash embedded development toolkit
And how does one develop under such a system? Post your programs and hope they get modded up to compile-threshold level?
DataSquid.net, a little about me.
Some version of this is already shipping inside gateway's AOL network access box. Yes, AOL access on top of linux.