Re: n00b wanting a pro PC
Posted: Fri May 01, 2009 6:13 pm
Then I'll just wait for Larrabee for my next machine build. 
Support and community hang-out spot for Ogre3D
https://forums.ogre3d.org/
Core 2 quads are a waste of time and money, it is not true that they outperform core 2 duo in almost any app, especially in Games, if you want a quad core go for the Core I7.Moohasha wrote: Duo core vs. Quad core: Right now I'm looking at the Intel Core 2 Duo E8500 3.16GHz which runs $189 on newegg, but the Intel Core 2 Quad Q9550 2.83GHz is $279. I know that a slower quad core is supposed to out perform a faster duo core in most cases, but what's the turning point, and is that always the case, or is it sorta app specific? And is the trade off worth the extra $90?
And why does that make you happy?jjp wrote:My bet is that Larrabee will be a disappointmentBut we'll see. …
Oh yeah Thanks for the welcoming. It has been almost eight month since last time I was around.xavier wrote:Victor, welcome back! We haven't had anyone to post BS and stir up the sh*t since you were here last.
I see that you are still a little conflicted with your politics.xavier wrote:You guys would make great U.S. Senators -- you're all awfully loose and fast with others' money.![]()
Yes of course, I have empirical evidence.xavier wrote:You of course have citations to back up this performance claim of yours, right? My experience, of course, is the exact opposite of your statement; my Q6600 outperforms the dual core next to it at work regularly.
I do it during compiles at work. After I quit smoking again it was easy to find the time to answer a post or two.VictorMoran wrote: I do not know how you do it, you being in the real world and all of that while we are in fantasy land.

Hell I forget which particular treachery of the public trust brought that up -- there are so many to choose from these days, with Congress spending like drunken sailors on shore leave and the administration almost falling over itself in its efforts to resurrect the ghosts of FDR, Johnson and Orwell, all at the same time, and before anyone notices...but we'd have to continue that in another topic.VictorMoran wrote:I see that you are still a little conflicted with your politics.xavier wrote:You guys would make great U.S. Senators -- you're all awfully loose and fast with others' money.![]()
![]()
I have found incredibuild actually to slow down builds, especially in cases where precompiled headers are involved (some of them can get pretty huge and transferring those around the grid often takes more time than just compiling the files locally). We use IB here too and since moving to a new codebase, I haven't found a need for it. Good header/dependency management plus judicious use of PCH almost eliminates the need for IB, IME. However, think about it -- VS will build 4 projects at a time instead of two on a quad core (vs a dual core) -- with enough memory in the machine, that will build twice as fast, no?VictorMoran wrote: We use an app called Xoreax Incredibuild, and it takes approximately from 10 to 15 minute to build the code.
Since the company no longer uses Dells, the new IT group set up another farm for the design team.
This one is made of 8 machines using ICore 7 with run a lower speed, and yet they compile the same code base in less than 7 minute.
A reasonable point, except that there's some uncertainty as to whether or not Core 2 Quad is at the end of its life -- they certainly are not pushing out the i7 like they were the Core 2 (I can only find 3 of them on newegg, and only one of them is "reasonably" priced).Another point was that if I were going to spend 2000 on a rig, I rather buy a system with future scalability and not one that is at the end of the development cycle.
The Core i7 *is* a better CPU, hands down -- but the questions raised so far have mostly been "is it *that much better*" -- by all accounts the i7 makes for a significantly more expensive machine, for which the performance in practice (I'm taking this anecdotally from those who claim to have tried both) simply doesn't justify the extra cost. If Intel would bring down the i7 cost and make them more available than they are, I'd agree with you that it is a no-brainer, but Intel's own distribution strategy seems to cast doubt on whether they actually are pushing it or not...The third, point I think that ICore 7 is better than the Quad cores is that they have a unified cache designed, but I am sure you know a lot more than I do about that.
the fourth, I believe it is Intel itself who say that Core I7s are better than the Q6600, and for more on that you can go to the Intel website and read the datasheet of the CPUs.
Well of course Intel says that, it's their latest chip which is about 2x-6x more expensive than the Core 2 Quads so they want you to buy it!VictorMoran wrote:the fourth, I believe it is Intel itself who say that Core I7s are better than the Q6600
Agreed. Not only are the CPUs much more expensive, but a mobo with a Core i7 slot is more expensive, as well as DDR3 RAM (which is the only RAM that Core i7 compatible mobos I've seen support). So the additional cost to getting a Core i7, from my experience, extends beyond the CPU iteslf.xavier wrote:The Core i7 *is* a better CPU, hands down -- but the questions raised so far have mostly been "is it *that much better*" -- by all accounts the i7 makes for a significantly more expensive machine
???I've actually found Vista (64, anyway) to be plenty stable...but annoying as all f*(&*ing hell...anytime (and I mean anytime) you want to do anything (including just running a goddamn program) you have to answer 20 questions about whether "you really wanted to do that?" And this is with UAC OFF...
Depends on what your expectations are.My bet is that Larrabee will be a disappointment
Wow, considering that this topic is related to picking computers parts, it sounds like your concerns with politics make you see analogies that seem like a stretch.xavier wrote: Hell I forget which particular treachery of the public trust brought that up -- there are so many to choose from these days, with Congress spending like drunken sailors on shore leave and the administration almost falling over itself in its efforts to resurrect the ghosts of FDR, Johnson and Orwell, all at the same time, and before anyone notices...but we'd have to continue that in another topic.![]()
Umm, It seems to me that you proved, either that Incredibuild is slow, or that my company is mediocre because we use incredibuild, or perhaps that we are mediocre for not using your vast knowledge in organizing makefiles to use precompiled headers.xavier wrote: I have found incredibuild actually to slow down builds, especially in cases where precompiled headers are involved (some of them can get pretty huge and transferring those around the grid often takes more time than just compiling the files locally). We use IB here too and since moving to a new codebase, I haven't found a need for it. Good header/dependency management plus judicious use of PCH almost eliminates the need for IB, IME. However, think about it -- VS will build 4 projects at a time instead of two on a quad core (vs a dual core) -- with enough memory in the machine, that will build twice as fast, no?
Umm I see, so what you are saying is: yes the core I7 is better, and it is the high end best performer according to Intel.The Core i7 *is* a better CPU, hands down -- but the questions raised so far have mostly been "is it *that much better*" -- by all accounts the i7 makes for a significantly more expensive machine, for which the performance in practice (I'm taking this anecdotally from those who claim to have tried both) simply doesn't justify the extra cost. If Intel would bring down the i7 cost and make them more available than they are, I'd agree with you that it is a no-brainer, but Intel's own distribution strategy seems to cast doubt on whether they actually are pushing it or not...
Or, perhaps, it's a light-hearted interjection of humour -- I'll let you figure out which it is.VictorMoran wrote:Wow, considering that this topic is related to picking computers parts, it sounds like your concerns with politics make you see analogies that seem like a stretch.xavier wrote: Hell I forget which particular treachery of the public trust brought that up -- there are so many to choose from these days, with Congress spending like drunken sailors on shore leave and the administration almost falling over itself in its efforts to resurrect the ghosts of FDR, Johnson and Orwell, all at the same time, and before anyone notices...but we'd have to continue that in another topic.![]()
I said none of that. You mentioned IB, I pointed out that in our case (with our previous codebase, horribly organized) that IB was almost no faster than a single quad core, and that with the codebase we since moved on to, that it was much faster on a single quad because the code org was much better and no grid jobs needed to be started (keeping the network out of it).Umm, It seems to me that you proved, either that Incredibuild is slow, or that my company is mediocre because we use incredibuild, or perhaps that we are mediocre for not using your vast knowledge in organizing makefiles to use precompiled headers.
It still fails to see how a xeon quad core can be much slower than a core I7 machine when they are running the exact same application. Do you have some knowledge Incredibuild is specifically optimized for Q6600?
Sorry if it wasn't clear to what that pertained.VictorMoran wrote:Core 2 quads are a waste of time and money, it is not true that they outperform core 2 duo in almost any app, especially in Games
I said no such thing. Simple observation provides that their product offerings for the i7 are (a) more expensive per CPU, (b) more expensive to set up, including the MB and RAM costs, and (c) that there are fewer of them than previous lines (which included typically a dozen price points at a time). From that, the only conclusions to draw are (a) Intel is (for some unknown reason) not pushing the i7 as hard as they did the Core/Core2, or (b) for some unknown reason, they have decided not to offer as many price points as in the past. I chose to conclude (a), since the Core 2 is still selling like hotcakes (and is probably cheaper for them to make, so margins are higher).Umm I see, so what you are saying is: yes the core I7 is better, and it is the high end best performer according to Intel.The Core i7 *is* a better CPU, hands down -- but the questions raised so far have mostly been "is it *that much better*" -- by all accounts the i7 makes for a significantly more expensive machine, for which the performance in practice (I'm taking this anecdotally from those who claim to have tried both) simply doesn't justify the extra cost. If Intel would bring down the i7 cost and make them more available than they are, I'd agree with you that it is a no-brainer, but Intel's own distribution strategy seems to cast doubt on whether they actually are pushing it or not...
But, because you have some intuition or some internal knowledge of what you call
“Intel's own distribution strategy”, the Icore 7 is not a better choice.
Thank you, I appreciate that.xavier wrote:Regarding the quad Xeon vs i7, in the case of building code it would be faster because there is more cache and memory access is much faster.
Considering that the issue in point was never formally introduced into the debate -- instead, you simply made it up on the fly in one of your replies -- there was no apology forthcoming. I offered apologies for not making sure it was clear to you which of your (mis)statements I was addressing with my quad/duo example, but that's not the same thing. Once again, in the above-quoted statement, you have introduced yet another tangential fallacy into the debate. But, you don't need to apologize for it. I'll simply direct you to the WIkipedia page on logical fallacies, since you tend to commit them more often than others, and let you figure out which ones apply to you. You can start with "strawman", "ad hominem" and "association".VictorMoran wrote:Thank you, I appreciate that.xavier wrote:Regarding the quad Xeon vs i7, in the case of building code it would be faster because there is more cache and memory access is much faster.
I knew that if I kept perseverance, one day you would accept that a person who is not in your real world, could say something right.
I still disagree with you in everything but I accept your apology.
Because there is some sort of hype around Larrabee. I am usually sceptical when a hype surrounds a product. So I can't help to feel a bit of schadenfreude when something gets hyped and then doesn't deliver. From a programmer's point of view the chip certainly is interesting but so are the chips from NVidia and AMD/ATI.VictorMoran wrote:And why does that make you happy?
For CPUs there is absolutely no difference in terms of stability and compatibility. None at all. For GPUs there are the differences between ATI & NVidia which I already mentioned. Apart from that they are equal in terms of stability and compatibility. For the best price/performance right now I'd get a GTX260-216.Moohasha wrote:I hear people saying that you get more bang for your buck with AMD over Intel, but can anyone aleviate my concerns about stability and compatibility with AMD? From what I've heard, more people have issues with AMD drivers than with Intel.
But that was spelt right!jjp wrote:So I can't help to feel a bit of schadenfreude when something gets hyped and then doesn't deliver.
Aha!jjp's profile wrote:Location: Karlsruhe, Germany
I've been running AMD CPUs for years now, with no issues (XP 32, XP x64 when it was RC, and now Linux). The only issues I've heard of with them are that there were some big bugs with the first Phenoms.Moohasha wrote: I hear people saying that you get more bang for your buck with AMD over Intel, but can anyone aleviate my concerns about stability and compatibility with AMD? From what I've heard, more people have issues with AMD drivers than with Intel.
The AMD Phemon IIs have great price/performance. Great if you want to make your money go a bit further. But Intel has the top end of the market and AMD's offerings fail to beat Intel's best. But like you said the Phemon IIs are great if you're on a budget or want to keep the price/perf ratio high. But just a word of warning, steer clear of the Phemon 1s, they are terrible compared to simmilarly priced Intel and AMD cpus)Owen wrote:I've been running AMD CPUs for years now, with no issues (XP 32, XP x64 when it was RC, and now Linux). The only issues I've heard of with them are that there were some big bugs with the first Phenoms.Moohasha wrote: I hear people saying that you get more bang for your buck with AMD over Intel, but can anyone aleviate my concerns about stability and compatibility with AMD? From what I've heard, more people have issues with AMD drivers than with Intel.
The new Phenom II X4s are looking very good, have a good price/performance ratio, and, if you're into it, overclock well. There are reports of the top end Phenom IIs overclocking to 4.2Ghz (from 3.6), which is mighty impressive - and would make it seem AMD has a lot more scalability available in the Phenom II micro architecture (I expect they're not releasing faster models at the moment for power consumption reasons - they're probably working on a lower power process revision before they unveil them - since at 3.6GHz they're already pushing 125W)
Moohasha wrote:I want to keep my new PC under $2,000, but still have something that will run well in a couple of years, so keep that in mind when responding.)[/b][/i]
Even though I know the answer, I'll ask anyway -- you have some figures to bear this out? I'm not doubting that it is, but I am wondering how your definition of "much faster" fits with reality.VictorMoran wrote: As a reference Quad Core Q9300 is much faster than the Q6600.
No wonder you seem so confused all the time, you can't even do basic math. I would love to hear how you get 55% from comparing a "regular core i7" to a Q9300? I can't even find a combination of numbers in the 3DMark06 results that comes out to 55%. We know you have a tendency to make things up out of whole cloth to back your arguments, but numbers don't lie...VictorMoran wrote: Here is a performance comparison chart for some Intel and AMD CPUs.
http://translate.google.com/translate?u ... l=el&tl=en
In a nutshell a regular core I7 is about 60% faster than an AMD Phenom X4 and about 55% faster that the top of the line Quad Core Q9300.
As a reference Quad Core Q9300 is much faster than the Q6600.
Wow dude, so much fervor, 25% faster, 36% faster, I eyeballed the figure but who is counting?xavier wrote:No wonder you seem so confused all the time, you can't even do basic math. I would love to hear how you get 55% from comparing a "regular core i7" to a Q9300? I can't even find a combination of numbers in the 3DMark06 results that comes out to 55%. We know you have a tendency to make things up out of whole cloth to back your arguments, but numbers don't lie...
And even assuming that your 55% figure is not a fantasy, it still begs the question: does such a number justify the additional cost? For example, a "regular core i7", which I take to mean "the 920", costs $280 at time of this posting, on Amazon, vs. a Q9300 on Amazon, which is $231.50. That comes to a 20% higher cost (280/231.50), for a 24% increase in performance (according to the 3DMark06 results in the link you provided). If by "regular" you meant "the 940", then for a 36% performance increase, you get to pay 2.46x the price (570/231); not exactly what I call a "justifiable expense".
....
[side note: in looking at CPU costs, clearly the economy-of-scale factor weighs in on Q6600 prices because they are more expensive -- far more in some places -- than the latest models in the Core2 Quad line. Wanted to point that out, since I've had mine for over a year -- and been perfectly happy -- and have had no reason to go looking at CPU prices in some time]
Apparently not you -- but inflating "facts" to bolster a weak case is about par for the course with you. And not "fervor", but "accuracy", a concept that obviously missed you. Finally, I'd get your eyes checked if you're that far off...VictorMoran wrote:Wow dude, so much fervor, 25% faster, 36% faster, I eyeballed the figure but who is counting?
Where in my previous post did I defend anything? You introduced the comparison between a "regular core i7" and the Q9300, and all of my points were made on that basis, except for the self-explanatory side comment at the end of the post.VictorMoran wrote:The passion you are displaying defending the legacy model Q6600 CPU will make anyone thinks that the Q6600 was actually the faster solution.
It was about $2000 in 2007, would you like the specs?God you must had paid a great deal of dough for your system.
Then you would continue to play the fool, a role you apparently wear proudly and like a glove. More than once in this thread I have stated that the Core i7 is a better and more performant design than the Core 2 Quad. The question, the one you apparently are ignoring, is "how much better" and "how much extra are we willing to pay for that performance". I've done the analysis for you in my previous post, and the numbers don't make the case for an i7.I still say, if I had $2000 to buy a system, I will be buying the top of the line Core I7, and you will be the one buying the Q6600.
You didn't learn much from that link I gave you on the strawman, did you? Here it is again, as you seem to have lost it:Do you think the Q6600 will make a comeback in the future?
Wow you are so feisty, I'd like that.xavier wrote:Apparently not you -- but inflating "facts" to bolster a weak case is about par for the course with you. And not "fervor", but "accuracy", a concept that obviously missed you. Finally, I'd get your eyes checked if you're that far off...
You see, I told you, now you start reusing the same material, it only means it is time to give up.xavier wrote:Wow, you didn't learn much from that link I gave you on the strawman, did you? Here it is again, since you seem to have lost it:
Here's what I see when I go to that link:VictorMoran wrote: I typed “$2000 Intel core I7 systems” in google and this is what I got:
http://www.anandtech.com/guides/showdoc.aspx?i=3510
I would be neither happy nor unhappy -- it is not I who am trying to make an invalid point. Once you understand what a "strawman" is you'll understand why your comments are so easy to deconstruct. And yet you keep coming back for more...Will it make you happy if I say the Q6600 is the better choice?