SUN in bed with Intel the monopolist
SUN is reported to be in bed with Intel. Am I surprised? Not a bit. I see through the current SUN leadership well before the rumor.
With nothing distinctive left, SUN is becoming just another assembler of offshelf commodity parts trying to grab a piece of the AMD64 pie. And SUN is in a worse positition than DELL, as DELL at least has a highly efficient supply chain.
Well, one could say SUN started off as a company built on commodity parts: free and open sourced BSD + Motorola CPUs was all that SUN had. That was a time when there were few players in the market and there were less than 100 good C programmers in the whole world.
Today the situation is different, any company can whip out a 8 way Opteron server using parts bought online. SUN's tiny volume of 10,000 server per quarter is simply not enough to win a discount from AMD. DELL sells more AMD servers per day than SUN does in a week.
SUN is desperately trying to steal a sale or two from DELL. The SUN+Intel deal is a not a deal between equals but between a little boy and big brother. Intel gives lip service for SUN's crappy software SUN isn't making a dime on, and SUN pays hard dollars for Intel bakery.
The current SUN leadership is driving SUN to the grave. Look at the software projects SUN tried to do in the past few years. Every single one of them was a total failure, and SUN evtually had to give it away for free -- no one wants to pay $ for them any way. SUN tried Star Office, Java Desktop (has nothing to do with Java), Java Enterprise, N1, $100 per employee subscription, J2ME..... They all end up the same. Today, SUN is still trying to beat the dead horse on its crappy software.
SUN's mating dance with Intel signifies the end of it. Before, SUN at least has the heart of the hard core techies. Now, SUN lost its moral essence and becomes just another also run. What SUN has is less than what Tyan or SuperMicro has. Tyan and SuperMicro can design better motherboards that hold AMD or Intel CPUs. AMD never dies, because AMD stands for something. SUN is not anything anymore, it's just another ordinary company heading to oblivion.
posted by Sharikou, Ph. D. @ 1/21/2007 07:17:00 PM
67 Comments:
LOL
The Pretender answered his own question. Why is SUN going INTEL?
Because "Today the situation is different, any company can whip out a 8 way Opteron server using parts bought online."
Thus to be leading edge you need the best from the best. That is INTEL Woodcrest. With K8L nowhere to be seen and AMD roadmap in shambles with low yielding 65nm and poor performing 65nm and huge design issues with Barcelona. The smart money has realized that INTEL is kicking A. 45nm Penrym samples are already at SUN and the results speak why Sun is going INTEL.
This blog is dead...............period.
AMD is fragged. No one wants to wait for AMD's promises of quad core processors when they can get Intel quad core processors now.
Intel is more than 2.5x faster in servers 2P than AMD. AMD is old junkyard 90nm technology.
Ahh.....so this is the reason you were putting Sun down.
http://news.com.com/Sun+servers+heading+back+to+Intel+chips/2100-1010_3-6151872.html
I think all AMD only server providers are now in the same situation what Dell was one year back ! Customers demanding choice !
Although I'm a supporter of AMD (yes, I love to support underdogs), after its 4x4 launch (which doesn't even take X2 3800), I don't think AMD has control of the desktop high end range in these period of times.
They should make a low cost 4x4 which we can plug in two X2 3800.
Sun wants to combat Dell!
So, I guess Intel may have a chance at cracking the SUN nut!
This has the look of a subtle but that more strategic move by Intel to damper headway/momentum made into the server market by rival AMD when it had the edge of the fastest processor on the market. I wonder if SUNW got ALL it should get out of negotiations if MSFT were doing the deal.
YHOO/GOOG search "invest_mavin"
"They should make a low cost 4x4 which we can plug in two X2 3800."
With the way AMD 2P+ works it is impossible. With Intel it could be possible, at least in theory. At least their old off-the-shelf P3's were able to work in 2P nicely, I'm not sure if anything else besides the socket has been changed in that regard.
AMD giving discounts is OK?
And Intel is the monopolist while Sun, up to now, has only offered AMD?
Please dont give proof that youre not really bright in the same news at least.
No need tp bash SUN over a marketing deal. SUN's survival is not based directly on AMD64, but on Solaris. The initially chose Opteron's because those are better chips. But...as you state, anyone can build they're own....
For SUN to survive they need Solaris to replace Linux. The Intel initiative helps that.
Don't be a fool. SUN knows which chips are best....
demand for opteron servers in the broad market must be really awful for someone like Sun to eat crow and adopt Xeons.
If anything its a sign how poor AMD's product offering are.
I'm sure Sun's shift towards Intel will accelerate it to BK. Too much demand is bad for any company, right?
The first servers from the partnership will begin arriving soon--in the first half of 2007--and Sun also will sell Xeon-based workstations, sources said.
In addition, Otellini plans to endorse the x86 version of Sun's Solaris operating system, elevating it to the mainstream status enjoyed by Windows and Linux. That means the chipmaker will devote engineering resources to validate the software and help with hardware support such as support for chip power management.
here:
----------------------------------
http://news.com.com/Sun+servers+heading+back+to+Intel+chips/2100-1010_3-6151872.html?tag=nefd.top
... walking on sunshine ...
----------------------------------
http://www.google.com/search?q=fkk+solaris
... so you know when boobs are real ...
The Sun is going down...eternal darkness awaits humanity!
I think all AMD only server providers are now in the same situation what Dell was one year back ! Customers demanding choice !
That number is about ZERO.
Thus to be leading edge you need the best from the best. That is INTEL Woodcrest. With K8L nowhere to be seen and AMD roadmap in shambles with low yielding 65nm and poor performing 65nm and huge design issues with Barcelona. The smart money has realized that INTEL is kicking A. 45nm Penrym samples are already at SUN and the results speak why Sun is going INTEL.
Woodrcest NOR Clovertown can beat Opteron in transactions (the main reason for buying a server).
All of you who can't see or appreciate the duopoly
MAKE ME SICK.
SUN will go after Intel for sure.
But in the end when Intel start rising their prices again (after dumping their crap processors P4 and PD) everyone will see their long term strategy wont be good. Maybe in the short term will be good for sure.
Unless....
The new K8 doesn’t look much good and having Ati doesn’t do nothing to SUN, however DELL and HP that’s another story because AMD now can give them the all packaged, even more than Intel can give them right now.
AMD from “weak” platforms as gone for complete platforms now, which even exceed what Intel is offering.
AMD BK Q4 07!
I'm sure Sun's shareholders have looked at AMD's poor results over the last year and are wondering just when Sun will dump AMD.
We can see that Intel beats AMD in just about every performance test and comes in with much lower power usage.
Sun, of course, makes the lowest performance blade server available. All because Sun uses AMD's dinosaur technology.
It is 100% unacceptable to Sun's shareholders for Sun to be completely reliant on a company, AMD, that has only paper launched processors in the last year. AMD has delivered no evidence that they can design, build, and ship a new microprocessor. The original Opteron design was stolen and the team that made it long gone. All that is left at AMD is a bunch of spin-meisters like Hector Ruiz.
Meanwhile Intel is shipping, in
volume, the highest performance desktop and server processors in the world, including quad-core across their desktop and server line.
In not too long it will be clear that AMD and their K-chips is nothing more than the K-mart of microprocessors (and GPUs): bottom end performance at loss leader pricing. Only the weakest and cheapest companies will go with AMD.
Intel & Sun's Solaris team up to make the Xeon and Solaris relevant; good luck, you both have years of work ahead of you. Untold billions have already been dumped into Xeon. I give it even odds because Sun’s engineers are among the best on the planet.
Here's a snippet from SUN's announcement slide:
Why Intel?
• Technical excellence (meaning AMD is severely behind)
• Processor roadmap (meaning AMD's K8L and future proucts are poor)
• Tremendous brand and reach (AMD's useless in marketing)
• Market expansion (Demand for the Opteron has saturated, if not shriking)
• Common vision (AMD's fusion and lack of vision going forward frightens SUN)
The frightening bit about this is that SUN knows what AMD has to offer. SUN would have been in close communication about the progress of K8L. We don't know what SUN knows, but all we know is that it isn't convince with what AMD have and made a decision to go Intel.
BK in when again?
The frightening bit about this is that SUN knows what AMD has to offer.
What did Michael Dell know? Why did he risk his billion dollar Intel biz and go on stage with Hector? AMD+DELL screws SUN.
What did Michael Dell know? Why did he risk his billion dollar Intel biz and go on stage with Hector? AMD+DELL screws SUN.
Nope. Wrong. Completely wrong.
Intel told Dell go into biz with AMD in order to suck the life out of it so Intel can push and stomp AMD more safely on other fronts and suck The Force out of it.
Meanwhile (I.) Intel develops the UltraSpeed32NanoWarpMeter technology, worships The Beast and steals loyal girlfriends from AMD fans, technicians etc. and initiates and converts them into DarkSide SexSlaves.
Meanwhile (II.) DarkSide worshipping Dell technicians analize and benchmark every AMD platform weakness, condense the info and feed it to Intel Kabala DarkSide Tech Team.
See how the plot unfolds?
You are all blinded by The Science.
----------------------------------
P.s.: btw. what exactly is Apple doing atm?
thekhalif
"Woodrcest NOR Clovertown can beat Opteron in transactions (the main reason for buying a server)."
Have you seen JBench results? You know that benchmark that simulates an enterprise-level Java application. Or did you mean that Opterons are better for running SpecFP?
"AMD+DELL screws SUN."
Might want to ammend that to AMD + Dell screws both AMD...
AMD's capacity - constrained.
AMD's inventory - increasing (due to reduced flexibility due to Dell's capacity requirements)
AMD's ASP's - decreasing due to one really big customer (Dell) getting volume pricing and selling lower range of AMD's products.
AMD's customers - pissed off due to AMD starving the channel.
Long term this may work out (if/when AMD has the capacity) but short term it will continue to plague AMD, especially when they could have used the increased profits to pay off the ATI deal and fab expansion. AMD had no need for increased demand when they announced the Dell deal - they were selling everything they could make...they simply shot their load too soon! (should've waited until ~Q2/Q3'07)
Sorry, JBench is another thing. I meant SPECjbb2005
The frightening bit about this is that SUN knows what AMD has to offer.
What did Michael Dell know? Why did he risk his billion dollar Intel biz and go on stage with Hector? AMD+DELL screws SUN.
Yeah, that sure did help Dell a whole lot, didn't it? I mean, from #1 to #2 in less than a year? Great move!
The rumors are that AMD will not be shipping quad core in volume until mid 2008.
If this is true, it explains why Sun is going with Intel.
It also would explain why AMD had to come up with "4x4".
In many ways, K8L is turning out to be AMD's "Itanic".
If your wife betray you, you must blame yourself. Ever and only YOU!
I have said it before and I will say it again.
Have any of you intel fanboys run a conroe at 99% cpu usage for any length of time?
Have any of you intel fanboys tried producing random number sets with the conroes?
When you run a conroe in basic code for a short while, it slows down to a crawl and never speeds back up without a complete reboot cycle.
Conroe does not produce random numbers.
Whether you use 2 cores or 4 cores or 8 cores of conroes you just multiply the problems many times over.
Intel has no place to address these problems maybe you fanboys can get intel to explain?
Mission and Vision
from here:
----------------------------------
http://www.awgp.org/gamma/MissionAndVision
----------------------------------
The basic tenets of the Mission include:
Rise of divinity in human beings and heavenly atmosphere on earth
Individual, familial and societal refinement
Healthy body, pure mind and civilized society
Atmavat sarvabhooteshu (all living beings are soulkins), Vashudhaiv kutumbkam (Entire earth is our family)
One nation, one language, one religion, one government
Everyone should get equal opportunity for self-growth irrespective of caste, creed, color, religion, region, language or sex
To achieve the above goals, Pandit Shriram Sharma Acharya (Gurudev) started the Global Movement for Transformation of Thoughts. Some of the tools and techniques of this movement include:
Three Means of Spiritual Awakening: (1) Upasana (Contemplating on divine virtues); (2) Sadhana (Practicing self-control for acquiring divine virtues); (3) Aradhana (Using acquired resources for the welfare of the society)
Propagation of Gayatri (collective wisdom) and Yagna (cooperative virtuous demeanor)
Yoga, Meditation and Alternate Healing
Huge volume of constructive thoughts on every aspect of human life written by Gurudev (equivalent to 100,000 pages of encyclopedia Britannica size books)
Three pronged approach: (1) Advertising campaign: To send the message of ‘Ushering in of New Era’ to the masses; (2) Creative campaign: To get active cooperation of intelligentsia in the process of new creation; (3) Combative campaign: To overcome the obstacles coming in the way of achievement of our aim.
Create addiction free society
Promote Vegetarianism
Work towards establishing a well balanced, pollution free ecosystem
Harmony, friendship, brotherhood, peace and prosperity of the whole universe
Promote Scientific Spirituality
Inculcate family and cultural values in younger generation via Multimedia and other state-of the art technology
Women awareness
All Free Thinkers and wishers for the Free World should aim their Minds towards the Noble Goal: to help AMD in their efforts to overcome obstacles and become the True Green Microdevices Provider - and to obtain the Nikola Tesla Wisdom and Vision, with the help and means of True Maharishi Effect:
----------------------------------
http://www.vedicknowledge.com/Maharishi_effect.html
Preteder you are a work of stink... LOL
"What did Michael Dell know? Why did he risk his billion dollar Intel biz and go on stage with Hector? AMD+DELL screws SUN. "
Michael knows nothing! He continues to have full confidence in Ken and look what has happened. HP has cleaned his clock. Growth is still stuck in the slow lane, HP is now #1. All this while embracing AMD. What a mess Dell is in. You heard it here, Dell is going the way of other famously once great but now BK companies; Wang, Data General, CDC... They are history. The most laughalble thing is the Pretender claimed AMD was going to save them. Look what happened, in the next month going INTEL is the death for SUn.. why not Dell... Duh...
Lets see what happened to Dell when they embraced AMD. Sales are down from 10.5 to 9.7M.. is that what you get when you start advertising AMD Turdions in your product? HP on the other hand is doing very well, Q4 are up 23.8% YoY rising from 9.6 to 11.9 Million units. Looks like AMD really helped Dell there Sharikou! Lets see how Apple did now that their line has converted to INTEL. Shipments up only 30% damm do I see a trend. Embrace INTEL core2 sales go up. Embrace AMD sales go south. But the PhD pretender will have another interpretation of course. He's the pretender remember!
Sun makes its money with software, services and hardware. Dell sells you a PC and hopes to make a few bucks.. broken strategy especially with AMD. HP got service, got ink, got profits, got growth.
Where are them Barcelonas, NOWHERE! Where are those screaming fast 65nm from AMD, nowhere! But wait we got Fusion coming to save us... Hector must be smoking some crack with his brothers.
This blog is dead...............period.
Amen to that! Sharikou, I think you have circled the wagons but you've run out of ammo and the braves are approaching. Better hold onto your scalp!
"Have any of you intel fanboys tried producing random number sets with the conroes?"
Well, you figured it out - I sit around all day generating random # sets to exercise my CPU, in fact I believe this is the #1 activity performed on all desktop and mobile computers
"Have any of you intel fanboys run a conroe at 99% cpu usage for any length of time?"
It's difficult as it completes the tasks so quickly, it never seems to stay at 99% for an extended length of time. Of course I'm not a "megatasker".
Solaris / Xeon (S&X) platform will be a niche market. Most enterprise will not load-up on MS-Vista for another year; how long do you think they’ll wait for S&X ?
This blog is dead...............period.
A dead blog does not speak.
Sharikou - why must you rev your blogs all the time? is your first post usually that poorly thought out?
At least have some integrity and mark the rev'd blogs with "update" or something to that effect...
As a "jouranlist" who has published numerous articles online - do those places (which coincidentally you've never identified) allow you to continuously update your articles after they've been published?
Here's a snippet from SUN's announcement slide:
Why Intel?
• Technical excellence (Intel's PowerPoint slides are better than AMD's)
• Processor roadmap (We dig Intel's groovy processor names.)
• Tremendous brand and reach (Intel's Mafia friends will lay-off if we play ball.)
• Market expansion (Intel's Mafia friends WON'T lay-off if we don't buy ALL that they dump on us!)
• Common vision (The blind leading the blind.)
This blog is dead...............period.
A dead blog does not speak.
No, a dead blog speaks but a dead blog doesn't SAY anything. Ergo, your blog is the prototypical dead blog.
"Have any of you intel fanboys run a conroe at 99% cpu usage for any length of time?"
It's difficult as it completes the tasks so quickly, it never seems to stay at 99% for an extended length of time. Of course I'm not a "megatasker".
An answer from an Intel fanboi for sure. When you stress test a system, you can make it run at 100% usage. Given the report last year from the Korean portal outfit, I bet Conroe systems run at 100% for some time will result in instability just as like what that Korean outfit ran into. Too bad they never reported on why that happened but one wonders why no report was ever made...
Sharikou - why must you rev your blogs all the time? is your first post usually that poorly thought out?
Sometime I first fire a shot, then write more based on readers' reaction.
"Have any of you intel fanboys run a conroe at 99% cpu usage for any length of time?"
Ran 2 instances of prime95 on my e6700 for a couple of days with no issues, actually now that I think about it I haven't rebooted my computer for several weeks. Sounds to me like someone is trying to spread FUD. Considering that they have sold something like 60 million C2D in the last six months don't you think that someone would of already caught this so called problem??? Hey I know maybe sharikou can go to best buy and test this out for you;)
"Sometime I first fire a shot, then write more based on readers' reaction."
Then why not annotate the changes?
An answer from an Intel fanboi for sure. When you stress test a system, you can make it run at 100% usage. Given the report last year from the Korean portal outfit, I bet Conroe systems run at 100% for some time will result in instability just as like what that Korean outfit ran into. Too bad they never reported on why that happened but one wonders why no report was ever made...
Wow. I wonder why the millions of Core 2 Duo based system owners haven't reported anything about that yet!?
"When you stress test a system, you can make it run at 100% usage. Given the report last year from the Korean portal outfit, I bet Conroe systems run at 100% for some time will result in instability just as like what that Korean outfit ran into. Too bad they never reported on why that happened but one wonders why no report was ever made..."
I have Seti@Home constantly running on both cores at all times on my e6300@3.1GHz. I also use that box for programming and enterntainment (S@H still runs while I do it).
It never hangs and there is zero instability. Uptimes are measured in weeks and machne is restarted only when needed. Though it might be from the fact I use Linux.
I have said it before and I will say it again.
Have any of you intel fanboys run a conroe at 99% cpu usage for any length of time?
Have any of you intel fanboys tried producing random number sets with the conroes?
When you run a conroe in basic code for a short while, it slows down to a crawl and never speeds back up without a complete reboot cycle.
Conroe does not produce random numbers.
Any links to this conspiracy or has it just started on this blog?
As for Solaris... in regards to the corporate space, I think it is better solution than Linux. But unfortunately Sun thinks it will thrive (but it's a dieing breed).
Sun signing on Intel only shows the duopoly in the making... just like AMD keeps signing Intel only outfits on a weekly basis.
While it may be surprising that Sun was moving towards Intel, given the firm's close connection with AMD, Schwartz had a good reason to call Otellini. According to Sun, 70% of Solaris installations on x86 systems run on Intel systems, not on AMD. Ignoring Intel meant ignoring 70% of its market. "That was a great motivation to work together," Schwartz said. Of course, the fact that Sun was able to convince Intel to market Solaris may have save the operating system's life: Seven million installations are considerable but the future of the software was uncertain: "There was the question if Solaris would suffer the same fate of other UNIXes," Schwartz said. "Now that issue is off the table."
For Intel, the agreement means that the company can work with a company that has found a new way of growing its server and workstation shipments, and has done so with AMD systems only. The fact that Sun is the fourth largest server manufacturer worldwide and that the company has gained credibility in the x86 market means that Intel can go after a substantial installed base of Sun AMD systems, even if AMD may retain unique selling points such as more consistent and easily upgradeable platforms.
...........
Industry analysts such as Dean McCarron from mercury Research, meanwhile mentioned that the cooperation between Intel and Sun "wasn't made for any one reason." Instead, Sun possibly saw a number of benefits from working together with what has been one of its worst enemies. Among the more significant reasons, the analyst believes that "similar to what Dell has done with AMD last year, Sun is now in a better position to negotiate better prices with AMD." McCarron mentioned that Sun's competitors in the x86 market, foremost Dell and HP are already offering AMD and Intel CPUs across the board - and Intel's quad-core Clovertown processors. Offering Intel processors will allow Sun to catch up with its rivals in the x86 market and maintain its accelerating pace.
from:
----------------------------------
http://www.tgdaily.com/2007/01/22/intel_sun_agreement/
----------------------------------
Quite possibly it's not only "Intel's technical excellence" but a coldblooded market-related decision, related to SUN's software/operating system.
BTW it it just me or is Paul Otellini brain-dead? If his presence on stage at the announcement is any indication I can fully reconcile the pathetic situation INTC is in now. Hopefully he was just having a bad day.
"What did Michael Dell know? Why did he risk his billion dollar Intel biz and go on stage with Hector? AMD+DELL screws SUN."
Michael Dell knows how to optimize his supply chain. How he could squeeze more out of AMD than from Intel for the same money.
Sharikou, when u fire a shot it usually the idiots that get hit.
- Scorp76
Maybe it was because of the "exploding" AMD chips?
http://www.theinquirer.net/default.aspx?article=27605
"Early Sun Fires with Windows, er caught fire"
Yes, it's an old story, but you know the saying- Burn me once, shame on you, burn me twice...
get a life ....
"Sun signing on Intel only shows the duopoly in the making... just like AMD keeps signing Intel only outfits on a weekly basis."
TO USE YOUR OWN WORDS - Any links to this conspiracy or has it just started on this blog?
WEEKLY signings of Intel only outfits?!? So for last 6 months that should be ~26 deals for Intel only outfits, no?
Look at that, AMD just wiped out all the profit they have ever made in their entire existance in just one quarter.
When's that Intel bankrupcy again?
40% Gross margin...ouch
Gotta love Dell! Who needs to sell the chips at higher price in the channel when we can give them to Dell at lower prices!
I run linux also and you have to examine the TOP numbers to analyze task loads.
In my tests seti does not use 100% of the cpu power but generating random number sets does.
It does not take long before a e6600 slows down to the speed of a 286.
Random number generators are used in the video games that most of the intel fanboys play.
A conroe is probably never giving you the game as intended by the programmer due to conroes inability to produce random numbers.
By the way, if you run a linux operating system and vexira virus scan at the same time the conroes just about drop dead.
I run debians linspire 5.0 v2 but I am sure problems will show up with conroes no matter what os your running especially with the random number problems.
I have run solaris which is a version of unix, but Linux is much more complete for desktop users.
If you leave your contact information I will be more than happy to send you the test procedures to duplicate my results.
If your intel, you should fix these problems before you do serious harm to this world.
If your AMD, keep up the good work your processors are doing just fine and test out very well, even the random numbers are right.
Yo Ph.D pretender!
AMD Opteron server shipments have slumped. Operating losses of 577 million dollars!
This compares to what, a profit of 1.5bn for Intel?
AMD lost over $500 million! AAHAHAHAHAHAHAHAHAAA!!!!!!!
http://www.businessweek.com/ap/financialnews/D8MR98B01.htm
So much for Intel going BK when AMD is up to it's eyeballs in debt and posting losses to the effect of half a billion.
So Doctor, got a comment on those fat 40% margins AMD is getting?
How about how AMD managed to wipe out all the profit it has ever made in a single quarter?
Who's BK in 2Q08?
Hey Shakira,
AMD lost over half a billion dollars. Say, who's going bankrupt again?
Wow. Over 6 hours since the results came out, and nary an update. Hell, you could hear a pin drop. Still trying to figure out how to spin this one positively?
BK gives out those little BK crowns for free you know...
"I run linux also and you have to examine the TOP numbers to analyze task loads."
I do.
"In my tests seti does not use 100% of the cpu power but generating random number sets does."
There must be something you are doing wrong then. Perhaps you have some other tasks running that take some of the CPU time. On my PC, CPU load rarely drops below 99%. When some other application besides seti needs CPU time it just takes whatever it needs and seti uses the rest.
On my P4 with HT Seti doesn't use all of the CPU when I'm in graphical interface but that is because I couldn't install ATI drivers properly and have to use software emulation. At home I have C2D with HW acceleration on NV and it works perfectly.
"It does not take long before a e6600 slows down to the speed of a 286."
Attatch the heat sink properly and/or clean it from dust.
"Random number generators are used in the video games that most of the intel fanboys play. A conroe is probably never giving you the game as intended by the programmer due to conroes inability to produce random numbers."
You are jokning, aren't you? I can write a program that generates hundreds of millions of random numbers per second. What program did you use to generate those numbers? glibc rand(), #echo $RANDOM or something else?
Using $RANDOM I can generate 1,000,000 random numbers in around 30s on 3GHz P4 (~33k per second). Using glibc's rand() I can generate 1,000,000,000 random numbers in 35 seconds (~28.5M per second). There are functions for generating random number sequences that are orders of magnitude faster than glibc's rand. So my question is, what the hell did you do to generate random numbers that slowly it might slow games down? Or are you saying that on C2D those two programs I wrote run significantly slower than on P4?
Normal games use at most couple of thousand random numbers per second and that is assuming high FPS (>100).
"If you leave your contact information I will be more than happy to send you the test procedures to duplicate my results."
You can view my email by clicking on my name of this post. Don't send emails to my msn address, I won't read them. I'm running 32bit Gentoo but it shouldn't matter in any case.
I hope you send those test cases soon and not chicken out.
ho ho
I tried both of your email connections but like contacting intel you cant get through.
Whos the chicken now.
You intel fanboys are all the same.
I still stand by my statement, intels conroe processors are not producing actual random numbers.
I really enjoy this blog its one of the few honest blogs on the internet.
"On my P4 with HT Seti doesn't use all of the CPU when I'm in graphical interface but that is because I couldn't install ATI drivers properly and have to use software emulation."
Wow, nice, I finally know what's P4's hyperthreading good for: Seti. ;-)
ATi has lousy Linux support, but its Windows drivers are just fine. It has better power management on Windows platform, too.
"Using glibc's rand() I can generate 1,000,000,000 random numbers in 35 seconds (~28.5M per second). There are functions for generating random number sequences that are orders of magnitude faster than glibc's rand."
I guess one important factor you purposely omitted here is the quality/strength of the random numbers. On Unix where there is support even /dev/urandom could generate more randomized pseudo-random values than rand().
The pseudo-random numbers the previous guy mentioned were probably for cryptogaphy. I'm sure you can't generate 1 billion cryptographically suitable PRNs on any PC in 35s.
anonymous
"I tried both of your email connections but like contacting intel you cant get through."
Then something has to be wrong at your end. That Email is correct or otherwise I couldn't write the posts here under my blogger name. You could also give me your contacts so I could contact you.
anonymous
"Whos the chicken now."
Well, since I know for a fact that my email works it must be you.
What answer did you get when you tried to email me? Perhaps you just made a typo somewhere.
anonymous
"I still stand by my statement, intels conroe processors are not producing actual random numbers."
Would you then describe how you reached to that conclusion? Last time you were talking it doesn't produce them fast enough, now you say those numbers aren't really random? Btw, have you got any idea how random numbers are generated?
I just ran the same C++ program on my 3.1GHz C2D I did a few days ago on P4. Result was that 1 billion random numbers were generated in 17 seconds, or about twice the speed P4 did it at the same clock. The program uses only one core.
edward
"Wow, nice, I finally know what's P4's hyperthreading good for: Seti. ;-)"
Actually, you are right. With running two clients in parallel with HT I gain around 10-15% performance compared to either without HT or running only one client.
edward
"ATi has lousy Linux support, but its Windows drivers are just fine. It has better power management on Windows platform, too."
It may be so but as I don't use Windows neither at home nor at work so it is kind of meaningless to me how good drivers they have under the OS I don't use.
edward
"I guess one important factor you purposely omitted here is the quality/strength of the random numbers."
standard rand() is random enough for most purpuses if you use it correctly (rand()%10 is not correct). If something better is needed then you can always use merseinne twister to get better numbers at higher speed.
edward
"On Unix where there is support even /dev/urandom could generate more randomized pseudo-random values than rand()."
Yes, it could. Problem is that it would be slow, on every machine, since it takes a (pseudo) file read to get anything from it. So far I haven't heard of anyone using it for anything meaningful when speed is needed.
edward
"I'm sure you can't generate 1 billion cryptographically suitable PRNs on any PC in 35s."
Tell me what library to use to generate those nubmers and I'll test it myself. Also the generation of the random nubmers is not the most demanding part in cryptography. In fact, it has very little impact on the speed.
Forgot to mention one thing:
edward
"The pseudo-random numbers the previous guy mentioned were probably for cryptogaphy"
Actually, (s)he was talking about games.
""The pseudo-random numbers the previous guy mentioned were probably for cryptogaphy"
Actually, (s)he was talking about games."
No, he talked about PRN by itself before he mentioned games. Then he made a leap of faith claiming the inability of making fast/good PRN would "probably" make the games wrong unsatisfiably. On on hand, there's no reason to believe games require crypto-strength PRN; on the other hand, this doesn't mean there isn't PRN that are computationally more challenging than glib rand().
Post a Comment
<< Home