Personal computing discussed

Moderators: farmpuma, just brew it!

 
rgreen83
Gerbil
Topic Author
Posts: 28
Joined: Thu May 12, 2005 1:58 pm

Now I understand how TheBank and OC-AMD feel.(rant)

Sat Aug 20, 2005 3:03 pm

I am posting here because I know you guys and gals have been very helpful and supportive and thats why I chose this team in the beginning. And also because I have seen you all struggle to understand why ppl leave as fast as they join. And lastly because I know some or all of you truly understand why we fold, and the desire to find answers as fast as possible.

The following is a post made by a user(7im, not a mod) at the folding-community forums:
Okay, enough with the violins. Enough with the points whine-n-cheese-fest. It's way past being very old. We heard you the first time, and the second, and third, & fourth, fifth, sixth, seventh, eighth, ninth, tenth, eleventh time, etc. Stanford has always said the mix of work units would be in constant change. I am NOT saying that your complaint is invalid, just really stale.

Start thinking up some solutions and quit your complaining. We are all here for solutions, right? Well start acting like it. Since you all can't seem to get the job done, here it is.

Fact. Intel compiler won't work on AMD chips. AMD compiler won't work on Intel chips. And it only applies to Windows clients.

Fine.

Option 1 (already suggested before): Split the Windows client. One for Intel users and one for AMD users. Still, that takes a year for Stanford to develop, test, and release, but it is a valid option. Take a long chill until then, but problem solved.

Option 2: Trade all of your AMD boxes in for Intel computers, and problem solved. What? Too drastic? Fine, try this one...

Option 3: Switch all of the AMD boxes to linux. The linux client doesn't check for CPUIDs. Problem solved. Still too drastic, fine....

Option 4: Run the F@H client and another DC project at the same time to make sure your fancy "under utilized, it's such a waste" processor gets worked to death. Problem solved.

Option 5: Don't run F@H at all. Problem solved.

Option 6: Have some patience. Stanford heard you the first time. They may not fix it tomorrow, but they always try to do the best they can to make the project work for everyone. This option is called "Do nothing, and wait for a new option." Problem solved, eventually.

Unless anyone has any better ideas, those are your options. Pick one.


And my original response:
7im wrote:
Option 5: Don't run F@H at all. Problem solved.


Well thanks, I think Ill take that one. You just dont get it, do you? I and others have already said that while points are nice and all it doesnt really matter. I am not just thinking of my boxes, but the tens or hundreds of thousands of others out there being underutilized also.

I understand this is not Pandegroup's fault and there my be nothing they can do about it, but while were all spinning our tires here, my grandpa is dying of colon, liver, and lung cancer because he worked in a dynamite factory all his life, and I already lost my cousin to leukemia when we were 13. Thats why I fold, points are fun but ultimately I dont give a crap about them. I just want to help this research progress as fast as possible.

How's that for your violin, 7im? If this is the attitude the users have in the project and the mods dont care either, then I think the 230 US Bank locations I was borging will be better spent helping ppl who understand what we're fighting for.

Everyone should take a minute to read this post on the sourceforge dev blogs http://blog.dev.sf.net/index.php?/archi ... bles..html .


When I said i would take option 5, i was being facetious, but 7im's assertions that users who are trying to do a good thing being "whiners" really upset me. This post was edited moments later by mod to read:
. . . EDIT by Mod:
Since you've already decided, we don't need to hear the rest of your rant . . . no matter what else you had to say.

There is no longer any way we can figure out how to help you.


I dont get it.

You can view the thread for full context herehttp://forum.folding-community.org/viewtopic.php?p=109801#109801.

Now I warned in the title this could be a rant so I hope it can stay as is, because i get nothing from ranting, i just would like the folks at pandegroup to be a little more understanding of their users wishes. Now I dont intend on leaving(my personal boxes will still contribute), because I still love the team here and believe in the cause, but like TheBank and OC-AMD, I dont feel comfortable asking my future father-in-law to run F@H at the 230 banks he runs until I feel that if a problem comes up as it did for TheBank and OC-AMD that Pandegroup will be there to help. This may be one of many possible reasons ppl dont hang around longer. Have others had similar experiences when venturing onto the Pandegroup's forums?
 
Maph
Gerbil Elite
Posts: 652
Joined: Sat Jan 22, 2005 11:31 am

Sat Aug 20, 2005 3:33 pm

ouch, that's pretty bad that the mod just edited out your whole post. I have been reading FC forum regularly, but that's the first time I have actually seen a mod do something drastic like that. I am glad you decide to continue folding on ur own box still even if you decide not to fold w/ those corp machines.
Image
 
bhtooefr
Silver subscriber
Lord High Gerbil
Posts: 8188
Joined: Mon Feb 16, 2004 11:20 am
Location: Newark, OH
Contact:

Sat Aug 20, 2005 4:11 pm

And with this, I think it's time to look for another DC project for TR to work on.

So, here's what we need:

It needs to serve the greater human interest. F@H is working on protein folding for disease research, so it definitely qualified. FAD, maybe?
The people behind the project need to listen to their users. I'm not so sure that F@H meets that requirement any more. UD is a definite no-no here.
Team competition is a must. FAD has it. And, if R3D moves to FAD, we can race them to Team Hard Cider, and do a full rematch of the doodyhead race :P
User competition is a REALLY good idea, as it allows for subteams. FAD's got them, as well.

Another perk: THINK (the FAD client) plays a LOT nicer with SMP than the F@h client. It also seems to allow you to pick the best WUs (called jobs on FAD) for your system, by letting you specify your CPU speed. There IS a (beta) Linux client. Unfortunately, the Windows client is not available in a console (ninja edited there...) version.

So, TR FAD? Anyone like it? I'll be putting this on -oneunit if you guys like the idea.
Image
 
KingFish
Gerbil Elite
Posts: 693
Joined: Sat Oct 30, 2004 10:20 am
Location: Soggy Bottom, Lousyana
Contact:

Sat Aug 20, 2005 5:03 pm

Wow, that's unlike them to edit posts like that. I don't see any reason that they would do that. That's also unlike 7im to lose his cool too. I've never seen that type of post out of him.

I'm sure it'll take some time to get the changes but hang in there, I'm sure they are coming sooner or later. It always seems that the more we want things that the longer it takes to get them. It makes it extra hard when loved ones are in a battle for something you are trying to help find a cure.

I've followed the AMD/Intel compiler "scandal" and can tell you that Intel definitely isn't playing fair and there isn't much that the PandeGroup can do short of writing their own compiler, which isn't likely due to the resources involved. They've already stated that their primary goal is research and want to get the most research done with what they have.

That said, with the number of different platforms out there with the processors numbering in the hundreds of thousands, it's almost impossible to wring every last MHz out of every last one of them. When you deal with processing on a scale like this, you'll have to make compromises to get the most realistically. Unfortunately that means that each processing platform may or may not get utilized to their fullest extent. Other posters at the F@H forums correctly point out that the pendulum has swung the way in favor of AMD before and now favors Intel.

With the qmd cores giving Intel such a boost it's likely that Stanford may rebenchmark the system and make adjustments just like it did before when the pendulum swung toward AMD. I have reasonable faith that Stanford is working to make the most out of the system and work out bugs even when it may not appear to do so. I've been vocal in the past about more feedback from them so we know what to expect so wild speculation isn't generated. It's always better to hear it from the horse's mouth anyway.

I'm whole-heartedly supportive of you folding and trying to find a cure for your loved ones but I don't see the lack of sse3 optimizations for your cpus as a barrier to you asking your future father in law for the borgs. We're all here to help our loved ones as well as yours. Any little bit helps, even if it's not as much as your processor is capable of giving.

If the ranting helps get it out of you, that's good too. 8)


KingFish
<a href=http://www.heatware.com/eval.php?id=25646>My Heatware</a>
<img src="http://chronos.dynup.net/Sigs/sig1.php?designer=Sparrow&sigid=3&user=KingFish">
AMD 3500+ | Asus A8V | 1 Gig Patriot TCCD Ram | Nvidia 6800 GT
 
idchafee
Gold subscriber
His Holy Gerbilness
Posts: 13965
Joined: Thu Sep 25, 2003 8:39 am
Location: Chicago, IL
Contact:

Sat Aug 20, 2005 5:12 pm

It seems to me that in recent months, Pandegroup has been....less responsive to user needs. Its almost like they don't realize that without us, there IS not FAH project. How would they get funding then?

And I don't understand that edit job at all. I can see someone being a Stanford apologist like that 7in guy. But speaking here as a mod, that was NOT an editable post.
YOU CAN RUPTURE SOMEONE'S SPLEEN WITH A WATER BALLOON!!!!
 
KingFish
Gerbil Elite
Posts: 693
Joined: Sat Oct 30, 2004 10:20 am
Location: Soggy Bottom, Lousyana
Contact:

Sat Aug 20, 2005 5:21 pm

I totally agree. That was not a post that was "out of bounds" or a violation of forum rules. It was simply a mod that did not agree with a post and decided to take it upon himself to impose his beliefs by wiping out an opposing opinion. It's simply a deplorable move by the mods over there.
<a href=http://www.heatware.com/eval.php?id=25646>My Heatware</a>

<img src="http://chronos.dynup.net/Sigs/sig1.php?designer=Sparrow&sigid=3&user=KingFish">

AMD 3500+ | Asus A8V | 1 Gig Patriot TCCD Ram | Nvidia 6800 GT
 
bhtooefr
Silver subscriber
Lord High Gerbil
Posts: 8188
Joined: Mon Feb 16, 2004 11:20 am
Location: Newark, OH
Contact:

Sat Aug 20, 2005 6:08 pm

I thought that someone posted some C++ code to fool the ICC into assuming that the processor manufacturer is "GenuineIntel", so that SSE2 and SSE3 is enabled on AMD processors that have it...

Why doesn't Stanford use that?
Image
 
KingFish
Gerbil Elite
Posts: 693
Joined: Sat Oct 30, 2004 10:20 am
Location: Soggy Bottom, Lousyana
Contact:

Sat Aug 20, 2005 6:43 pm

That would be a cool hack. Intel would probably quickly sue Stanford if they implemented that. I would imagine end users would have to do it on their own.

Anyone else know about this?
<a href=http://www.heatware.com/eval.php?id=25646>My Heatware</a>

<img src="http://chronos.dynup.net/Sigs/sig1.php?designer=Sparrow&sigid=3&user=KingFish">

AMD 3500+ | Asus A8V | 1 Gig Patriot TCCD Ram | Nvidia 6800 GT
 
bhtooefr
Silver subscriber
Lord High Gerbil
Posts: 8188
Joined: Mon Feb 16, 2004 11:20 am
Location: Newark, OH
Contact:

Sat Aug 20, 2005 6:56 pm

Here's a link to a Slashdot post referenced by TR's article on the compiler screwing over AMD: http://yro.slashdot.org/comments.pl?sid ... d=13042922

It forces a sane memcpy algorithm on non-P4 CPUs (as I understand it, anything that supports MMX).

Also, here's a usenet post (linked from the comments on that TR article) that mentions the fooling of ICC into running SSE2 on ALL supported CPUs: http://groups.google.ca/group/comp.arch ... 8&oe=UTF-8

As for making end users make the mod, that's REALLY hard to ask of them. See, first, they'd have to buy ICC (VCPP isn't as fast, and GCC is dog slow). Then, they'd have to have source to the cores that Stanford is using, which is something that Stanford specifically won't do (it's QMD that's the real issue, and QMD is fully closed-source).
Image
 
Maph
Gerbil Elite
Posts: 652
Joined: Sat Jan 22, 2005 11:31 am

Sat Aug 20, 2005 8:51 pm

idchafee wrote:
It seems to me that in recent months, Pandegroup has been....less responsive to user needs. Its almost like they don't realize that without us, there IS not FAH project. How would they get funding then?

And I don't understand that edit job at all. I can see someone being a Stanford apologist like that 7in guy. But speaking here as a mod, that was NOT an editable post.


Dr. Pande said in a post during the whole oc-amd/how strike aftermath that in the future he will be less inclined to change the way they run the project due to user complaint/discontent, but so far it doesnt look like anyone representing the actual group has said anything in that thread. I thought they have always said that Mods are not part of the Stanford/Pandegroup, but I could be wrong. Who on TR has a voice over there? I know alot of us do read/post there but most of us don't exactly have much of a weight there...
Image
 
astrotech66
Gerbil Elite
Posts: 711
Joined: Thu Jun 17, 2004 10:48 am
Location: San Antonio, TX

Sat Aug 20, 2005 9:11 pm

bhtooefr wrote:
So, TR FAD? Anyone like it? I'll be putting this on -oneunit if you guys like the idea.


I'm interested ... I've been doing f@h for awhile and am maybe getting a little burnt out on it. Plus, I use only AMD machines, so I'm getting a little tired of the whole compiler issue. Are you going to start a TR FAD team?
 
idchafee
Gold subscriber
His Holy Gerbilness
Posts: 13965
Joined: Thu Sep 25, 2003 8:39 am
Location: Chicago, IL
Contact:

Sat Aug 20, 2005 9:15 pm

Maph wrote:
Dr. Pande said in a post during the whole oc-amd/how strike aftermath that in the future he will be less inclined to change the way they run the project due to user complaint/discontent


Then Dr. Pande can expect to not have a FAH project very long.
YOU CAN RUPTURE SOMEONE'S SPLEEN WITH A WATER BALLOON!!!!
 
bhtooefr
Silver subscriber
Lord High Gerbil
Posts: 8188
Joined: Mon Feb 16, 2004 11:20 am
Location: Newark, OH
Contact:

Sat Aug 20, 2005 9:49 pm

You want me to lead the team? I can get an e-mail sent TONIGHT for FAD to add us as a team.
Image
 
leor
Maximum Gerbil
Posts: 4782
Joined: Wed Dec 11, 2002 6:34 pm
Location: NYC
Contact:

Sat Aug 20, 2005 9:59 pm

let's all chill out and grab a beer.

if pande and those jumped everytime someone got pissy they would get flooded with complaints to change this and change that and would spend all their time hopping from one foot to the other.

i agree there are problems, but as a person who is running this thing at home, all my friends and family members, and like 5 small offices, i'm able to manage just fine.

i get the whole idea of being incensed at the prospect that they don't care about our needs, but i dont think that is the case here.

what they're doing with OC-AMD's request does from a organizational standpoint set a bad precedent, and i'm sure if someone comes up with a valid point in the future they will address it. they're not just going to boneheadedly ignore people for the sake of ignoring them, but by the same token they can't look like a bunch of bitches that will just lay down for anyone that complains.
 
Flying Fox
Gerbil God
Posts: 25402
Joined: Mon May 24, 2004 2:19 am
Contact:

Sat Aug 20, 2005 10:17 pm

I see this from a couple of different perspectives.

A lot of people do see points as the only thing as the competition really heats up, but tend to forget what is actually being done here. Sure the ppd is lower, but you still complete the same number of WUs with your machines and Standford is still getting the same amount of data. And as the project grows to include ever more people every day, there will just be more noises and the variety of people just increases. They will be shouting for things in different directions and it is just hard to cater to all of them. And when they get ignored because they feel like they have every god given right to be heard and acted on, they shout like a wild animal. If we ever forget what is truly being done here (research not points), I don't think people will be complaining too much. Some really do behave badly in the forums, for that I would put on Ed Stroligo's face and say "grow up". But I do think editing the post is not exactly correct.

UD has been having its troubles lately too and a lot of people just hopped onto the forums and complain about them leaving and such. And the old RC5 and SETI has been losing people. It's just that when a project is running long people become bored and obssessed. They just lose interest and will be looking for other things to play. To maintain the hard work persistently is tough, we want the project to be completed, not dies midway through.

I would say we calm down a bit before really thinking about leaving. We are doing solid work here (and there are more papers released than I originally expected) so it is still worthwhile. And as some on that thread mentioned, the pendulum will swing again.

My last point: if all those people are leaving and we stick around, wouldn't that give us the chance at the top spot? :P

Ninja edit: OC-AMD was asking for an option to be put in to limit RAM usage, eventually it is implemented. That's something that Stanford can do. But the compiler issue is really out of their hands, so that's just unfortunate. While I sympathize the situation, I don't see much that they can do to rectify it, given that the source code is closed and such.
 
bhtooefr
Silver subscriber
Lord High Gerbil
Posts: 8188
Joined: Mon Feb 16, 2004 11:20 am
Location: Newark, OH
Contact:

Sat Aug 20, 2005 10:28 pm

They could read my second link, which tells how to fool the compiler into allowing SSE2 and SSE3 boosts on AMD64 chips (and possibly P6 chips *cough*Pentium M*cough*) that support it. The thing about the source being closed that I said was that THEIR source is closed. They can implement the fix in their own code.
Image
 
Flying Fox
Gerbil God
Posts: 25402
Joined: Mon May 24, 2004 2:19 am
Contact:

Sat Aug 20, 2005 10:40 pm

It will violate the EULA of Intel's compiler. That's why their hands are tied.

Corporate computers will not run code that are questionable. IT managers don't want trouble. It will be a bigger mess.
 
KingFish
Gerbil Elite
Posts: 693
Joined: Sat Oct 30, 2004 10:20 am
Location: Soggy Bottom, Lousyana
Contact:

Sat Aug 20, 2005 10:50 pm

I agree with grayfox :lol: and leor. Let's not jump off into a knee-jerk reaction. Maybe we all should just step back and take a breather until we settle down a bit. I see no compelling reason to jump ship at this point. I see the project as quite honorable and producing good solid research. No distributed computing project will be perfect given the sheer number of computers involved and the many different platforms supported.

Hey Maph, you have a url where Dr. Pande makes those statements in the forums? I haven't heard or seen him say that myself, it interests me though.


KingFish
<a href=http://www.heatware.com/eval.php?id=25646>My Heatware</a>

<img src="http://chronos.dynup.net/Sigs/sig1.php?designer=Sparrow&sigid=3&user=KingFish">

AMD 3500+ | Asus A8V | 1 Gig Patriot TCCD Ram | Nvidia 6800 GT
 
bhtooefr
Silver subscriber
Lord High Gerbil
Posts: 8188
Joined: Mon Feb 16, 2004 11:20 am
Location: Newark, OH
Contact:

Sat Aug 20, 2005 10:54 pm

Talking about this sentence? (yes, I found a copy of the EULA)

ICC 9.0 EULA wrote:
You may NOT: ... (v) reverse engineer, decompile, or disassemble the Materials


The thing that I linked to was first, made using a disassembly of the RESULTING BINARY (read: not Intel code), and second, it was for ICC 8.0 (which means that things might have changed).

FWIW, I DID find that ICC is free for NC use...
Image
 
Maph
Gerbil Elite
Posts: 652
Joined: Sat Jan 22, 2005 11:31 am

Sat Aug 20, 2005 10:54 pm

KingFish wrote:
I agree with grayfox :lol: and leor. Let's not jump off into a knee-jerk reaction. Maybe we all should just step back and take a breather until we settle down a bit. I see no compelling reason to jump ship at this point. I see the project as quite honorable and producing good solid research. No distributed computing project will be perfect given the sheer number of computers involved and the many different platforms supported.

Hey Maph, you have a url where Dr. Pande makes those statements in the forums? I haven't heard or seen him say that myself, it interests me though.


KingFish


i can't tell u which thread it was in exactly, since there were several threads dealing w/ that siutation at the time...I wills ee if i can find it tho...will post it if i do...
Image
 
KingFish
Gerbil Elite
Posts: 693
Joined: Sat Oct 30, 2004 10:20 am
Location: Soggy Bottom, Lousyana
Contact:

Sat Aug 20, 2005 11:08 pm

k, thanks 8)
It seems a bit out of character for him as he stated that he takes donator relations quite seriously and put in quite a few posts at many different forums trying to mend fences when the mutiny commenced. I can imagine that he was quite weary from that long weekend but I can't really picture him saying that. It's just appears to be a bit ironic.
<a href=http://www.heatware.com/eval.php?id=25646>My Heatware</a>

<img src="http://chronos.dynup.net/Sigs/sig1.php?designer=Sparrow&sigid=3&user=KingFish">

AMD 3500+ | Asus A8V | 1 Gig Patriot TCCD Ram | Nvidia 6800 GT
 
Maph
Gerbil Elite
Posts: 652
Joined: Sat Jan 22, 2005 11:31 am

Sat Aug 20, 2005 11:14 pm

KingFish wrote:
k, thanks 8)
It seems a bit out of character for him as he stated that he takes donator relations quite seriously and put in quite a few posts at many different forums trying to mend fences when the mutiny commenced. I can imagine that he was quite weary from that long weekend but I can't really picture him saying that. It's just appears to be a bit ironic.


okey found it, page 8 of the QMD suspended thread in general FAH(i was looking at the wrong thread earlier and had a hard time finding it...end up clicking on Dr. Pande's name and look through all his posts recnetly...) ugh

http://forum.folding-community.org/viewtopic.php?t=12756&postdays=0&postorder=asc&start=105
Thanks for the feedback. People have responded to your point #2 and your answer to point #3 is itself pretty much all there is to say for now. For point #1, it's important to keep in mind that this memory limiting feature is really only a shadow of what it would have been in v6 and was quickly put into v5 to try to handle an imperative need.

However, there will be negative consequences of releasing this in v5, without the additional needed code to be in v6. While it's good for us to be responsive to donors needs, this may delay v6 a bit as we'll have to re-architect some aspects.

So in a sense, this was a trade for short term benefit to some donors vs long term benefit to the project in general. In the future, I may likely be less willing to make such tradeoffs.


not exactly the way i put it, but that's from the feeling i got out of the post reading it at that time...which was from a month ago...
Image
 
bhtooefr
Silver subscriber
Lord High Gerbil
Posts: 8188
Joined: Mon Feb 16, 2004 11:20 am
Location: Newark, OH
Contact:

Sat Aug 20, 2005 11:40 pm

Still, it comes off to me as a sort of "those ungrateful folders, we had to get this done before we felt like it, because they'll leave otherwise"...
Image
 
dragongoddess
Gerbil Elite
Posts: 536
Joined: Fri Aug 20, 2004 8:26 am

Sat Aug 20, 2005 11:47 pm

Hate to break some peoples hearts but 7im is right. This whole farce of CPU's not being fully utilized is nothing more than someone crying about points. Get over it guys. This is serious basic medical research. That comes first before anything else. So quit crying about points. There was a time when wu's had point values of less than 1 point and a 1 pointer was the wu to have.

I also want to know why someone would want to pull members of TR away from F@H. Is there some hidden agenda here that I'm missing.
<img src=http://dcsig.liquidninjas.com/store/dcsig_1268_3526.jpg>
 
Maph
Gerbil Elite
Posts: 652
Joined: Sat Jan 22, 2005 11:31 am

Sun Aug 21, 2005 12:04 am

Ya the compiler thing is nothing anyone can do about, and its not worth losing sleep over. big corp folders have their reasons to want someway to limit ram usage and if it affects the normal operation of their systems they have to decide wether contributing to the science is enough to justify running folding. I think for us normal folders it is not a big deal, tho any improvement to the client is a good thing for everyone, and getting big points is always good in our races/challenges against other folders/teams :)
Image
 
Flying Fox
Gerbil God
Posts: 25402
Joined: Mon May 24, 2004 2:19 am
Contact:

Sun Aug 21, 2005 12:15 am

bhtooefr wrote:
Still, it comes off to me as a sort of "those ungrateful folders, we had to get this done before we felt like it, because they'll leave otherwise"...
(empahsis mine)

That's most likely the reason why they rushed putting it in 5.04, because OC-AMD did have the influence and respect in the community. They did pay attention. However, I don't think the first part of the above speculation is true.

And as OC-AMD said, it was a difficult decision, and this is exactly what we are seeing what was feared to happen. On one hand the corp computers are hurting, adding the feature will produce immediate benefit. On the other, the feature has been planned anyway and OC-AMD was fully aware of that because of his heavy involvement. He does stay in the forums to help and keeps folding on his own machines.

Normal folders who don't follow this closely or lack the understanding of what is going on and what is involved will no doubt take it the wrong way, but the cause is too big a beast to let this controversy stop attacking it. Being a relatively matured forum TR is, we shouldn't rush to a decision.

New projects may be nice and shiny for the geeks in all of us, but not finishing anything that we have set out to do seem a bit... er... don't know which word to use.... :roll:
 
dragongoddess
Gerbil Elite
Posts: 536
Joined: Fri Aug 20, 2004 8:26 am

Sun Aug 21, 2005 12:41 am

Flying Fox wrote:
New projects may be nice and shiny for the geeks in all of us, but not finishing anything that we have set out to do seem a bit... er... don't know which word to use.... :roll:


How about childish.
<img src=http://dcsig.liquidninjas.com/store/dcsig_1268_3526.jpg>
 
leor
Maximum Gerbil
Posts: 4782
Joined: Wed Dec 11, 2002 6:34 pm
Location: NYC
Contact:

Sun Aug 21, 2005 12:51 am

dragongoddess wrote:
How about childish.


i like her

see sig |
|
|
|
|
v
 
Forge
Lord High Gerbil
Posts: 8214
Joined: Wed Dec 26, 2001 7:00 pm
Location: SouthEast PA

Sun Aug 21, 2005 1:21 am

GCC doesn't give two shirts what you're doing with the output, optimizes for everyone and their brother equally, and is fully peer-reviewed and open for patches/additions. It runs on and compiles for Win32, Linux32/64, Solaris32/64, *BSD, and more.

If Pande wanted a solution, there are solutions to be had.

Since they don't, I'm off to shut down my running clients and remove them from startup. They've been a constant low-grade drain on my box's responsiveness and ram, and I don't slow my box for aholes. Pande qualifies IMNSHO.

If there's an apology/unedit for rgreen83, I might reconsider, but otherwise I'm out and Pande group can suck it. I do free work for very few people/groups, and none of those disregard and insult me and other workers and get away with it.


Crap, and I was edging towards the team TR 200s with my machine alone...
Behemoth - Plex server
Galactica - FreeNAS
Pegasus - R710 ESX host
fsociety - MBP 15
Rocinante - E5530 testing machine
Saratoga, Ticonderoga, BunkerHill - the Hydra desktop PC-thing.
 
bhtooefr
Silver subscriber
Lord High Gerbil
Posts: 8188
Joined: Mon Feb 16, 2004 11:20 am
Location: Newark, OH
Contact:

Sun Aug 21, 2005 6:56 am

Forge: Here's the deal. GCC, while it optimizes equally, doesn't optimize much. GCC with full optimizations would most likely be slower than ICC on a K8.

That's why the fix (WHICH DOES NOT VIOLATE THE EULA FOR ICC) should be implemented to fool ICC into running the optimized code on a K8...

That said, I agree with you, Forge, on FAH. I'm switching to -oneunit, b/c I want to get this unit done.

I'm not looking at it as a points issue, I'm looking at it as a "do I want these guys running their code on my system, heating up my CPU" issue. I'm not so sure that I do.
Image

Who is online

Users browsing this forum: No registered users and 3 guests