Slashdot Mirror
Pix of The Crusoe Chips
A reader pointed us over to Transmetazone, a Transmeta Weblog, (Hey, why not!). They've got some cool pictures of the Crusoe chips - to scale. Take a gander at the TM5400 & the TM3120.
← Back to Stories (view on slashdot.org)
"Slightly difficult to make out are the 474 balls (small black dots) which enable its connections."
So I guess you could say that the Crusoe processor has a lot of balls.
(No, I'm not proud of this post. It's not my finest work. Post 2.0 will be better, I promise.)
Save the whales. Feed the hungry. Free the mallocs.
The push-pin is a much better investment in computing power than a crusoe. For the price of the Crusoe you can buy several hundred push-pins which can be used to poke or stab your way into getting someone ELSE to buy you the Crusoe. Or two.
marotti.com
On the website they said they didn't know what those four holes around the microprocessor were for. They're for mounting a heatsink. The slot one celerons use the same architecture for mounting their heatsinks.
So why exactly are they called "balls"? I don't remember reading anything about balls being used to connect the chips to things on the Transmeta site. Is it some kind of ball & socket joint, or just a way to create a smaller area to solder things to the chip?
They've gotta be the smallest chips I've ever seen. If I ever buy a machine with a Crusoe in it, I'll be sure never to open the case for fear of the chip dropping out and being lost between the fibers of my carpet.... Yeah, I know the chip would be held in pretty well, but that would just be my luck.
Eruantalon
Eruantalon
The Annals of Middle-earth
With it being this small do you all think I could build a decent Beowulf in a cigar box?
threadeds blog
The Clueless weekly Award goes to the author of that page.
So microprocessors are small. Why do you think they are called "micro"? Does he think an AMD K6-2 is big as a pick-up truck? Or Pentiuns are the size of a brick?
Showing a processor without it's encapsulement (sp?) is a dirty marketing trick. Anyway, size does not matter.
[]'s Carlos Cardoso - Becoming a brazilian ProBlogger, typo by typo
Looks can be deceiving, but it's cooler than PIII (temperature-wise :)
Is the "typical Win9x application" running on those blank-screen laptops "reboot.exe"?
--
Here is the result of your Slashdot Purity Test.
Linux MAPI Server!
http://www.openone.com/software/MailOne/
(Exchange Migration HOWTO coming soon)
Well, the crusoe is indeed a very tiny looking chip in both of its forms. However, it should be remembered that these chips are only every this size as a bare unit that they are now; unless the unit is to be 'hard-wired' to the mobo (removing the possibility of future upgrade) the chip will have to be mounted on a 'plug'.
:)
It can be noticed that the PII/PIII and AMD K6 / Athlons all are mounted on 'plugs' which are then inserted into a socket on the mobo. Have you ever broken up an old CPU and discovered the size of the actual chip: typically no much more than 2x2cm. (I submit here that maybe the PIII 'plug' is a little on the gigantic side considering the chip within
Not to worry tho: if what you are looking for is a hugely powerful miniture PDA, a hard-wired crusoe is definately a top option, as for a laptop or even desktop; a plug option has to be the way to promote sales with the option of upgradability
Ripping an new rectum in the fabric of spacetime.
It doesn't seem likely that a dual configuration is possible. The northbridge of a machine's chipset would be responsible for allowing MPS. The Crusoe has a northbridge built into it, so unless they specifically designed it to allow multiple processors this would be unlikely. This may be something they decide to add in to a future Crusoe processor.
-
So, mount a heat sink hey ...
1) The versions shown were restricted in some way: when running flat out they need cooling?
2) The motherboard is for a different processor: so the pinouts match?
3) They put some holes in the board to give people with too much freetime something to think about.
threadeds blog
Not to sound sour but this Transmeta hype is beginning to get a bit galling, we've seen a lot of pictures of chips, we've seen suggestions of users, but in terms of cold hard reality its been a bit on the thin side. Every dribble from Transmeta is slapped onto Slashdot as if this is the defacto winner. Come on guys get a bit of perspective, they appear to be a great company with some great people, but many companies have failed even though their ideas seemed the coolest or the best.
I wish Transmeta all the success they can earn, but as with the ignoring of the Windows2000 launch, this is exactly the sort of attitude that
An Eye for an Eye will make the whole world blind - Gandhi
Whelp, I pulled out a 386 chip I had lying around (A80386DX-20) and removed the casing.
:-)
It's 1.6 x 1.6 cm... and that came out in 1985.
I'd imagine that they managed to make theirs smaller over the course of 15 years...
46. The Hobo smiles, his eyes glaze over, and he burps. "Beware the man who has lived longer than the Wasteland."
If the TMS5400 was to be used for laptops and the TMS3120 was to be used for handhelds... why is the TMS3120 smaller than the TMS5400? Does this make sense? Are they fabbed on the same process? If not, then OK, I understand, but if they are why would a better performing chip be smaller than a slower chip? I would think added complexity and increased performance would make a chip bigger. Either the sizes are wrong or they are fabbed at two separate feature sizes. (I dunno maybe .22 and .18, or .18 and .15?)
Also if one is bigger than the other there will probably lower yield associated and thus higher cost. So a slower chip would be more expensive than a faster one? Again, it doesn't make sense.
-
-alex-
lyx me.
rats - too many layers of metal to see what's really on there :-) but that's life for modern chips - Now days with 4-5 layers of metal we put the power on the top thicker layers which occludes the core that we're used to seeing in pretty die photos. This also makes microprobing during debug a royal pain :-(
Actually there's two sets of balls in question. (btw -many chips are that well equipped)
The first is the solder bumps (~200um diameter) that connect the chip to the carrier. As was previously mentioned, this technique was developed at IBM probably in the 60's/70's but wirebonding (~30um gold wire stiched a la sewing machine) sufficed until signal requirement growth outpaced the density that could be packed on the edge of the chip. This "flip chip" or C4 (Controlled-Collapse-Chip-Connect -trademark etc IBM) spreads the I/O + power connections out over the entire surface of the chip.
FYI- there are two primary methods of applying these solder bumps, neither of which involve liquid solder. One is to place the whole wafer in a liquid bath and plate solder onto the conductive spots on the chip. The second is to put the wafer in a vacuum chamber with a mask on it leaving holes where solder is to be applied. Then the metals to be deposited are evaporated, filling the holes in the mask, which is eventually removed.
The second set of balls are the ~1mm dia. solder balls that attach the carrier to the circuit board. These are actually pre-formed (like BB's) and are sifted into the correct spots in a tray before being melted to the carrier.
naturally you might ask- why not just mount the chip straight to the board? Well, there's many reasons, but the main two are: 1-board building is cheapest at ~1mm dimensions, and 2-the thermal expansion of the chip/board are quite mismatched, which eventually causes something to flex and break. (the carrier mitigates this somewhat)
I know that money isn't a big topic around here, but everyone wants to know when Transmeta is going to IPO.. I was looking at the names of the top people in Transmeta and the Chief Financial Officer would be the most likely to have the details and plans for an IPO. So I decided to research the dude...
"Dan E. Steimle"
Went to google and the first thing that came up was a lawsuit against the guy buy stock holders of a company called "Hybrid Networks". They are charging him and others with Securities fraud. The defendents demanded a trial by jury.
This makes me nervous about Transmeta. Why would Dave Ditzel hire someone like this (with a grey background, and possibly pending litigation for SEC fraud)????
Scarry... Anyone have any thoughts?
How long did it take between Palm having the Color enabled OS, all the off-the-shelf components selected, and the emulator written before they finally actually released the Palm IIIc?
In that case, they were done, or working with already in-use components, and they were also up against a competitor that had a "feature"[1] that they lacked, and thus needed go get that advantage.
Hardware production takes time. You usually have to wait a minimum of six months before you can actually buy the stuff shown off from the assembly lines from the Taiwanese trade shows.
I use Windows because a multi-head display is a "killer-app" for me. I've been waiting a year for XFree86 4.0. I'm patient... I haven't started loudly proclaiming that they are vaporware: I know that the delays occur, and when it ships, I'll buy it.
[1] Incidently, is it just me, or do others think that color is *not* worth having battery life go from "2-3 months" to "2 weeks"? Especially when I use my Palm so heavily I run through batteries every 3 weeks anyway?
--
Evan
"$30 for the One True Ring. $10 each additional ring!" -- JRR "Bob" Tolkien
BGAs are pretty much permanent installations. I believe that there are clamshell sockets available for this kind of thing though. Bigger and bulkier, but you can just open up the socket and replace the chip if you need to.
Overall, BGAs aren't all they are cracked up to be. The production processes are very finicky. It's hard to get all the solder balls to form acceptable welds, and the welds aren't particularly durable in any case. Flexing a board with a BGA chip can very easily loosen or break a solder connection, causing intermittent or permanent failure, respectively.
If you want to get rid of the bathwater, you've got to throw out a few babies.
I use Windows because a multi-head display is a "killer-app" for me. I've been waiting a year for XFree86 4.0.
Snag yesterday's snapshot. I've been using XF86 snaps since 3.9.15, and this one resolves each and every issue I had with dual-head in the earlier releases. I'm currently running dual Matrox G200SD/16M, but I have used heterogeneous card setups like the S3 ViRGE PCI/AGP anything combo so common under Win98/NT.
Oh, yeah.. Snag the source if you have enough bandwidth.. You'll only have to 'patch'n'go' when 4.0 rolls out in a month or two.
.sig: Now legally binding!
Amen, :)
On the other hand, considering how long the batteries on WinCE devices last, if that was the only thing holding you to the platform, the PalmIIIc's battery life should seem like nirvana
This space for rent. All reasonable inquiries will be entertained at proprietors discretion.
Yeah, and we all knew about the transmeta release, but we got several news articles on it when it happened.
/. isn't just about notifying people "look here's somethingyou didn't know", it's about discussing about it too.
most people also don't care about a kernel 2.3.14159 release on the same day as windows 2000. which is more important?