Slashdot Mirror
Compaq Helps You "Test Drive" Linux and Unix
Ron Rangel writes " This website gives you quick, open, online access to simply run your applications on Linux. Or at least thats what they say." I tried it and it was kind of fun. You sign up (moderately intrusive questions), then Telnet (no ssh) into servers running several flavors of Unix and several Linux distros on different (Compaq) hardware platforms. Want to play with SuSE Linux on an Alpha (or whatever)? Here's your chance!
Without disputing your overall assertion of security, I would like to take issue with your claim that "most of the cracker tools refused to compile/run properly due to its 64-bit architecture". To me, this sounds like "NT is secure because Unix cracking tools don't work". You're assuming that no one is capable of doing anything more than compiling sources that they don't understand. They can port the cracking tools, or write their own.
Gates' Law: Every 18 months, the speed of software halves.
Their SuSE box is somewhat more usefully set up than the others. The RedHat box seems to not allow outgoing network connections. Bah. The tru64 box doesn't have any kind of sane bash environment set up and has no X clients installed.
The SuSE box allows outgoing network connections (I have an xterm open on it) and even has ssh installed!
By the way, be careful typing those IP addresses... I got the wrong one by mistake and in place of a login I got something like "This is not a public machine. Your attempt to connect has been logged."
This is the newest encarnation of the world-beating clustering technology that everybody is trying to copy. Shared everything, multiple OS instances in the same box, dynamically reassignable memory/CPU between instances, seemless clustering. Read the Galaxy overview here. VMS has been doing clustering since 1985 better than any Unix does it today and it's improved a lot since then. Yet somehow, OpenVMS gets the rap as being outmoded.
Disclosure statement: Yes, I am a Compaq employee. No, I'm not speaking for Compaq.
I was playing along with a small encryption program I have written (Idea implementation) which takes suitable amount of time on my home machine.
./encrypt_cc ./encrypt_ccc
:)
The results are fascinating:
Encrypting a file named "data" without any compiler optimization switches took:
[ph2ph@spe85 ~]$ ls -la data
-rw-r--r-- 1 ph2ph nis 76592295 Sep 24 21:46 data
[ph2ph@spe85 ~]$ time
72.125u 1.414s 4:31.80 27.0% 0+0k 0+0io 76pf+0w
[ph2ph@spe85 ~]$ time
26.203u 1.169s 0:42.64 64.1% 0+0k 0+0io 78pf+0w
As you can see, things are pretty freaky.. 42 seconds with ccc (*their* compiler) versus 4.31 minutes with gcc. I could've compiled with -02 however that optimizes code about 2 times only...
Anyone have a comment on that? Hmm.. btw this happened on the RedHat machine. I'm still to test on the Tru64
-- offtopic response --
Hey, its a small time operation. I have very little users, and virtually zero load. And I can take my dsl up to 2/3 the speed of a T1, which more than ample for a sizeable number of users. For the bots and small web pages on my box, its more than plenty at the moment.
I have yet to recive a complaint from any user to date.
----------------------------------------------
I don't really mind double posts on
I was there a few weeks ago, on the dual DS-20 Redhat machine, and even ended up in a talk session with the root user there for well over an hour (wish I could find his email so I could get his name...).
When I first talked with him, I mentioned that I was surprised Slashdot hadn't caught wind of it, though Freshmeat had. I then got to read the root user begging me not to post it here, as he wasn't nearly prepared for it =) Due to the results I'll mention in following, I quickly lost interest in it.
The machine was fairly impressive, but had obvious problems. Compiling was practically impossible as it crashed errors reminicent of a severely overheated CPU. Either that, or they had some major code problems. I only tried out gcc compiling kernels, and never did try out Compaq's compiler.
For years now, the Alpha's have reigned supreme as the absolute best CPU around, despite it's age. If only Compaq or one of the licensee companies, such as Samsung (I believe Intel and a few others have rights to produce the chips as well) would just pump out the chips and eat the losses for a year or so, the x86 architecture would be obsolete in probably 2-4 years. According to the man I spoke to, Compaq was about to start doing exactly that with the less expensive DS-10 chips.
Also, he said they would have a contest soon to win a free DS-10 based machine. I know Compaq has done some publicity on it, but basically it's a contest to write software to best demonstrate the abilities of an Alpha processor versus an x86. Or, at least, that was that I was told a few weeks ago.
Still dreaming of having a quad-81364/2Ghz machine on my desk in a couple years.... [drool]...
"Compaq ... supporting the Linux open source community since 1994"
Eh? How have they been serciving the open source community since 1994?
At any rate... I think this is a rairly good thing. At the very least, it gives average folks at home who may thing that *nix is too complicated a chance to see it. And this may be the only chance for a whole bunch of people, who are afraid to commit to using it, and who don't have access to it anywhere else (work, school, etc.).
--
Matt Singerman
Matt Singerman
http://matt.vegan.net/
I guess it would get really interesting, if they could provide even more operating systems from the *nix sector. How about comparing several Linux distributions to each other plus the free *BSDs and several commercial *nix variants?
Bjoern
New Linux cracking contest. Do I get to keep it if I win?
http://www.testdrive.compaq.com/_.survey.shtml
I ran a homemade benchmark on the RH EV6 and on my Pentium Pro 200 and my Celeron 366.
;) ) and then uses a random number generator to select and increment the cells in a pattern that biases it heavily toward the front, creating a curve of memory activity much like many applications. Note that there is no floating point code in the benchmark. The results are below along with my celeron 366 and pentium pro 200 (256K cache): ;)
The benchmark allocates a 1 MB array (256K x 32-bit unsigned int (no, I didn't use "long" on the alpha
ev6@500MHz: 7.23seconds cpu
celeron366: 8.49seconds cpu
pentiumpro: 13.67seconds cpu
The ev6 beat my humble 366c but I'm afraid my friend's 500 MHz Celeron beats it soundly. I don't remember the numbers he told me yesterday, but I remember it beating my 366 by 25% or so.
Geeky modern art T-shirts
Same benchmark as above but using a 64megabyte array. BTW, there is also a fair amount of string copying in the benchmark. With the 64MB benchmark, memory access speed starts to dominate the results more. The Celeron 366 won this round, which doesn't say a lot for the memory architecture of the EV6 (192.233.54.85). The Celeron, for those not familiar with it, has a full core speed 128K level 2 cache on the chip and accesses memory via a 66-MHz memory bus. (some people overclock the chip and use faster memory -- but not me ;)
Celeron366: 17.3 seconds CPU
EV6@500MHz: 19.3 seconds CPU
PentiumPro: 27.0 seconds CPU (200MHz, 256Kcache)
Geeky modern art T-shirts
I have knowen about this for quite a time and decided not to post it due to the /. effect. =(
This is geared twards application developers to get things done, unfortunately its been posted to slashdot now, so I guess every script kiddie and there leet gr00p are hacking away at it right now.
Crap. Now where am I going to compile new kernels at.
symetrix. We are building a religion, a limited edition.
What??? Obsolete compiler? Obsolete kernel? What are you talking about?
While these are not the absolute latest version, they are fairly recent (RH6 uses 2.2.5 and egcs-1.1.2). Have you checked the uptime on these boxes? (I don't have access...). 2.2.11 and 2.2.12 came out only in the last couple of months or so. It is highly likely that the boxes haven't been restarted since then...(why should they be?)
AFAIK, there are no distros that come with SMP compiled into it out of the box. So, someone HAD to compile this SMP kernel. This was not a corporate maintained computer! Someone took the time to install the kernel.
The compiler is a fairly standard one (not the latest and great, but still it is within 3-4 months old). gcc-2.95 is out, but the differences are not groundbreaking. There are still tons of people using gcc 2.6 and lower out there. A lot of the changes have to do with Intel optimizations, not Alpha optimizations...
This is not supposed to be a development box, but rather a test box to showcase the hardware and a guide to help port some of the code to run on these screamers.
The X clients are disabled for security reasons. You do not need X in order to test compilation and porting. The X libraries have already been ported to Alpha and work rather well. This should not be the focus of any porting effort. The X libraries are just that - the provide a standard API to which to code.
It goes without mention why they can not see any box outside of dec.com. Duh. Security 101 will tell you why...
BTW, a dual Celeron will die in comparison to a duel Alpha. This is of course assuming an optimized compiler and code on both sides. The Alpha is 64-bit with tons of cache. The Celeron is a 32-bit processor with hardly any cache. There isn't a chance in hell that the crippled Celeron can keep up.
Crawl back from whence you came!
Justin
P.S. I know this is pure flamebait, but this person is a complete yutz!
Mu. P.S. The address you see is real. =)
Ok, moderator types... the top-ranked comment on just about every story I've read today has been pumped up on "funny." And most of it's not that funny.
:)
Let's exercise those "insightful" and "informative" buttons once in a while, eh?
According to the HBS Press 1996 book " The Internet Strategy Handbook ", DEC (now owned by Compaq) was doing something similar already in late 1992 with their Alpha "Test Drive" program.
I'm not sure if they had anything similar after that, but it is nice to see them feature Linux this way.
--
Midgard Project - Open Source CMS
- A.P.
--
"One World, one Web, one Program" - Microsoft promotional ad
"Remember when the U.S. had a drug problem, and then we declared a War On Drugs, and now you can't buy drugs anymore?"
"Byte-manipulation isn't one of Alpha's strong points..."
;)
;)
Sorry... I should have said "memory copying" since I use memcpy(). As for Byte-manipulation not being a strong point of Alpha, well, isn't the point of a general-purpose CPU/memory subsystem benchmark to uncover these kind of things?
"If you want a real idea of memory bandwidth... use...Stream..."
Nope, I wasn't trying to assess memory bandwidth. By "memory access speed" I was referring to memory latency, a critical variable to system performance.
"...benchmark using the applications and data that are important to you."
heheh. Yeah, sure. The reason synthetic benchmarks and "application conglomeration" benchmarks exist is largely because of: 1) It's frequently damned hard to benchmark systems using the real application, complete with users, data, and all that. 2) Some computers are used in a very general purpose way. I do all manner of work on my little linux boxes, for example, and that pales in comparison to the wide-ranging use of some of the larger systems where I work.
"...it's irresponsible to post benchmark results without making the benchmark available...".
Dang. You've got me there. I wonder if we can get the Imark benchmarks from Intel, btw, to see how they've been biasing them toward the newer instructions as they introduce extended instruction sets...
Anyway, my page contains a link to the code for the 1 megabyte version of the benchmark in question.
Geeky modern art T-shirts
Of course it's not geard towards newbies. Compaq is trying to show nerds how fast their systems are, not newbies how easy Linux is. They sell boxes, not OSes; their target is technical people who already know UNIX but want to be amazed by the power of the Alpha. [And I must admit, it's pretty fast, though I haven't really tried too hard yet.]
"...gcc on alpha is not terribly efficient."
;)
Yes, you're right on this one. I compiled with the Compaq compiler with -O and got much better results: 2.2 seconds CPU time.
Looks like the lowly Celeron is soundly beaten, after all.
BTW, the code is posted on my web page, listed in a comment above.
Geeky modern art T-shirts
Same code, same options on the RH Intel box, results: 5.47 seconds CPU, less than half as fast as the EV6 500 at 2.2 (using Compaq's compiler).
Code link is on page listed in above comment.
Geeky modern art T-shirts
Talked to root too and got some additional info:
Although I think this site is a really great idea, I think they could do alot more to make Linux more palatable for those people who have "wanted to work with Linux, but never got arount to it." For instance, instead of just allowing them access to the command line, have screen shots and help documentation pop up in the browser, have an index, a tutorial, streaming video. Many people shy away from Linux due to the amount of time and resources (i.e. lost computer if unsuccessfully installed), this seems to me like the first step towards a pain-free linux trial.
Seriously - I hope they firewalled those boxes so they can't be used as springboards to attack other sites. This could quickly turn into a publicity-stunt-gone-bad.
--
[me@spe82 /]$ cat /proc/cpuinfo
cpu : Alpha
cpu model : EV56
cpu variation : 7
cpu revision : 0
cpu serial number :
system type : Rawhide
system variation : Tincup
system revision : 0
system serial number : NI82904549
cycle frequency [Hz] : 531914893
timer frequency [Hz] : 1200.00
page size [bytes] : 8192
phys. address bits : 40
max. addr. space # : 127
BogoMIPS : 671.08
kernel unaligned acc : 0 (pc=0,va=0)
user unaligned acc : 75 (pc=120002aa0,va=120106134)
platform string : AlphaServer 1200 5/533 4MB
CPUs probed 2 active 2 map 0x3 IPIs 3509024
Don't throw your computer out the window, throw the Windows out of your computer!
I've been using this little cluster for a while to port appliations to the various flavors they offer.
They get to show off their hardware, which is rather nice, and I get ported apps.
These aren't very loaded machines tho, If you want to do anything more then porting or benchmarking, on them look elsewhere.
- Adam L. Beberg - The Cosm Project - http://www.mithral.com/
After you play with it a while. (actually i think the idea is that you are suppose to upload, build/port and run your own software). Anyway when you are done, fill out the questionare and they will send you a Linux license plate (way cool), a redhat cap (red), and a little toy compaq car (gave to my kid :)).
The difference between Canada and the USA is that in Canada healthcare is a right and gun ownership is a privilege.
login: icepick
;)
Password:
Welcome to RedHat Alpha Linux 6.0 Land!
This is a Dual AlphaServer 1200
Please do not run any "RC5/SETI type" processes
If you do I will dis-user your account
That's too bad.
It's about time compaq does something for linux. I'm pretty tired of them simply resting on the laurels of digital engineers. I think this can have a very positive effect for fence-sitting would be linux users. Its even less work than vmware for people who want to try out linux. Of course compaq has a long way to go if they want to be able to claim they are supporting linux to the degree that sgi and ibm have been. of course if digital were still alive, they'd have the crown. long live dec.
-earl
I recently built a dual celeron box(abit bp6, 366 clerons, etc.) and When I installed RedHat 6.0, it installed a SMP kernel by default.
Sometime during the installation, it sees if you have more then one CPU or not, and installs what kernel it needs to.
I just assumed that other distos had SMP default kernels too. Or is Red Hat 6.0 the only one?
When I first checked this out it was my feeling that this was intended for corporations who are evaluating possibilities to port their software to a different platform.
Say they've got an awesome app on Unixware. But they heard that a 64bit platform will greatly enhance performance? How do they know how hard it will be to port the app, and then whether there'll be any benefit? They'd have to buy the hardware and tools, and everything just for a chance to evaluate what the platform will do for them. A lot of people are obviously turned away by the inherent risks. This takes that risk away by providing the tools for free, and even providing "contracts" to get your app ported.
I don't think their intent was for 50,000 /.'ers to get a free account to "mess" around in :-) I surely doubt this was for newbies either, hopefully if they are experienced enough to have software to run on it, they've passwd the newbie stage.
-Brent--
Compile time is a very poor indicator of speed
between different architectures. It's only a good
indicator within the same OS, architecture, and
compiler version.
An Alpha is a RISC CPU, and requires a lot more
intelligence from the compiler, especially in
areas like instruction reordering, than a CISC
CPU such as the ia32 family (i.e. your K6).
Also, GCC has been running much longer on ia32
than it has on Alpha, and therefore the speed
of the compiler itself has been better optimized.
The degree of optimization, simplification,
instruction analysis, and instruction reordering
needed on Alpha (versus x86) explains the long
compilation times.
ERROR: Null
If more vendors did this, there would be no legitimate reason to crack anymore. All of us hackers could make sure our stuff works cross-platform, and explore and learn about different systems, and law enforcement can stop bugging us, and hunt down the crackers and the script kiddies. :)
pb Reply or e-mail; don't vaguely moderate.
Your Test-drive account will remain active for 30 days from your registration date.
That's just super, the idea is there, and a really good idea it is, but only allowing 30 days to use this kind of defeats the purpose. I don't have anything I want to test right this minute, but I know in the future I might, deffinetly would be a plus if this was some form of permanent account.
Even if they were to say, "accounts that are inactive for 30 days are deleted" that would be fine, but to provide only a 30 day usability period is not exactly the greatest.