Intel's playing catch-up

Page 6 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.


This isn't quite accurate, While it is true that CAS 4 DDR2 800 has the same latency as CAS 5 DDR2 1000, the DDR2 1000 still has a higher bandwidth. Latency isn't all that matters - bandwidth is important too (in fact, in many applications, the bandwidth is actually more important than the latency).
 


You can whine all you wish. The evidence is in the forum. If you choose to not seek it yourself, then yes, I will lable you as lazy. While it is your right to be lazy, do you really believe that I have either the time or the inclination to suckle information to you, just so you dont have to bother yourself with looking for it? If so, you are sorely mistaken. I will not, and now because in your offense at being labeled 'lazy' you have made a point of avoiding the search and attempting to paint me as the villain to your "innocence", I have made it a point to not provide you any help in locating that information. Most interestingly, several of the individuals who have participated in this thread were participants in the threads where the information you refuse to look for resides, yet they have not provided you with links either. I wonder why that is?



I strongly recommend you contemplate the underlying logic of such statements before you post them. Had you thought for even the briefest of moments about your concept, you would probably arrived at the following conclusion:

Logic dictates that if, as you propose, I had some sort of "personal crusade" against thunderman, I would simply ban him and delete his posts.

Frankly, few would notice or mourn his absence, yet obviously I have not banned him, nor teletubbized him or babbled him, and he is still free to post, As such, your attempt to paint me as a being on some "personal crusade" fails, and the logic is so fundementally flawed that I must post a picture to emphasize your failure.

Similarly, if I had any such inclinations, I could ban you or anyone else I choose. I have not nor will I. In fact, in my time as a moderator, I have only ever permenently banned one regular, and temperarily banned 3. This does not include the "stranger" spamers or advertards. Soemthing else to consider when you are casting aspersions about 'personal crusades', or 'steering threads'

fail.jpg


As far as your 'call on me', allow me to make this perfectly clear. You chose, of your own free will to become involved. You could have chose to remain unnivolved, but you did not. As such, you are as responsible for the interaction between us as thunderman and myself are responsible for the interaction between ourselves. Once again, this is an open forum. No one can expect to post and not have their posts responded to (though some do) yet you seem to expect something for nothing, in that apparently you beleive you can engage in the exchange, get what you want, and not have to do any of the legwork yourself. Heres another one that I am not going to provide links to, that you are going to have to find yourself, meaning it wont happen: When Fuads blog first hit the net, thunderman did infact state that 'Nehalem won't overclock', not "....Nehalem won't overclock unless you buy a top end extreme edition..." as you say in one of your posts. Why do I bring this up? Simply because it is the crux of the warning to thunderman. He has stated that Nehalem won't overclock before, and been proven wrong. He has since modified his statement, yet is still wrong, and he knows it. Once again, had you bothered reading beyond this thread, you would know that, but like so many others, you chose to jump in the middle of a thread and pick a side without researching first. That, in and of it self is niether uncommon nor unforgivable, yet you have been told that 'there is more' than what is in this thread. Failing to take that hint to go find the information is unforgivable and taints you.

In regards to publically chastizing thunderman, there is an old adage "Praise in public, punish in private". These are very wise words, yet there are times, in extreme cases when a 'public example' needs be made as warning to all. As in the case of someone who consistantly, persistantly misbehaves. This was one of those cases. If you dont like it, I honestly dont care, nor do I have to. As long as the reaction I intend is the one I gain, which is the gradual return of the forum to normal behavior, then I will be quite content regardless of any feathers I ruffled to achieve that result. In my protracted absense, the CPU forum has once again deteriorated into a 50 gallon drum of napalm. Now that I am back, (at least for the time being until work calls me away again) the unbelievabley excessive flaming will stop. A little flaming every once in a while is nothing, and if the parties involved do so from a perch of mutually good natured fun, or honest debate then all is well and good, but the state the forum is in now is unacceptable. You can choose to be part of the solution, or part of the problem and try to 'call me out' , but ultimately it matters little as the excessive flaming will stop and your input will have no effect.

In regards to "steering this thread where I want", you are absolutely, positively correct. I will steer this thread where I want, which is to the facts, and away from the fanboy lies.



LOL, and much as it pains me to provide you any kind of link or evidence, Im afraid I must in this case, but only to prove that you dont read. Like several other of our more notorious posters (whom, by the way, I have also not banned) you apparently read only far enough to see that which you want to see, then rush out to the proverbial pulpit to preach your sermon:

In this case, you latched on to a single acronym: FUD.

Had you actually read, you would have seen many other terms including: “trolling”, “shilll”, “spreading lies” . The first two terms of these terms were modified by the (really try to pay attention now, as this is critically improtant) adjective “deliberately” The third term (again, try to pay attention) was modified by another adjective, “willfully”.





Now, in regards to the TERMS OF SERVICE, as I stated, not the Terms Of Use as you claim, I must provide a quote:

hate propaganda or hate mongering, swearing, or fraudulent material or activity

In deleberatly trolling, shilling, willfully lying or posting FUD, Thundermans post consitutes the violation of the TOS to which I was refering, specifically fraudulent activity.

fraud•u•lent
–adjective
1. characterized by, involving, or proceeding from fraud, as actions, enterprise, methods, or gains: a fraudulent scheme to evade taxes.
2. given to or using fraud, as a person; cheating; dishonest.

Random House Unabridged Dictionary, © Random House, Inc. 2006.

fraud
1. deceit, trickery, sharp practice, or breach of confidence, perpetrated for profit or to gain some unfair or dishonest advantage.
2. a particular instance of such deceit or trickery: mail fraud; election frauds.
3. any deception, trickery, or humbug: That diet book is a fraud and a waste of time.
4. a person who makes deceitful pretenses; sham; poseur.

Random House Unabridged Dictionary, © Random House, Inc. 2006.

And do you know what the funny thing is? In your rush to demonstrate you are not lazy, and to attempt to discredit me, the very tidbit from the terms of service that you chose to use (which you failed to cite, by the way) is the very same line which contains the specific term of service that thunderman is violating. Do you have any concept of the level of irony you achieved? Epic is far too shallow a word to cover it. You read the line, but you did not “read”. Totally, unabashedly, classic.

Now, as far as that particular line in the terms of service and swearing, Goddam is not swearing. Perhaps in your culture/religion it is, but as this particular THG site is intended for the US, and the US is by law non-secular, Goddam is not 'swearing'. And should you choose to counter with the "well it offends me" argument, I will simply counter with "OK, it offends you, therefore I must remove it. In the same vein, you offend me, therefore you must remove yourself. Do so now"








The meaning of "family forum" is that the site is open to everyone above the age of 13. In other words, as I have stated so often, "open to everyone" and not the personal domain of anyone individual or group. The meaning of "not responsible" is to protect BoM from legal action, not to “make it ok to offend children”. And yes, clearly you are not interested in reading or studying. You do not 'read' meaning while you may consume the written word, you can not or choose not to comprehend. Were you interested in 'reading' you would have understood that for yourself without having to have it explained to you.

You also seeming have another comprehension difficulty. You seem to feel it is my responsibility to ‘wet nurse’ you. You are wrong. My responsibility as a moderator is to maintain some semblence of decorum within the forum while at the same time keeping restrictions on freedom of expression to an absolute bare minimum or less. It is not to spoon feed every kiddie who is too lazy to search for information. It is that simple.

Now, because this thread has deteriorated into yet another annoying and pointlessly derailed flame fest, I am going to lock it. However, in the intereset of avoiding more annoying whining from you , I am going to give you the last word. So, go ahead and post your final response. Take your time, make sure you are happy with it. When I see that you have posted it, the thread will have the locking it so richly deserves. But fear not, I will not delete the thread. This way everyone can partake of your uncontested words of wisdom.


*douses thread in petrol and ignites*

fire or decapitation kills pretty much all know things, lets see what it will take to kill this thread.

It will be locked soon enough. Don’t you have a wingding waiting for you to pleasure him???




Then THG would be like a couple of other sites (who shall remain un-named) which are little more than thinly disguised blogs..

 
This thread was going places, now it went nowhere.

Close it, nothing good is to come now. Just exacerbated suppositions and endless ranting. Im sad, it started nicely tough.
 
Another pointless thread ...

It was fun though I think Keith is a bit bruised ...

Keith ... you have really big nads mate ... I respect that !!

Plentry of bandwidth too ... erk ...

Still ... it is nice to see we all don't agree on many things.

That's what make up the free world.

Just don't start wearing bandanas and toting any guns or anything ok guys??

I am running some smoothness tests on my rigs here ... lol.

I have noticed the Q is rougher at idle than the Phenom ... but I tweaked the engine management system so she advances a bit quicker now ...

I figure something funny might be a good end ... sorry for the epic fail there ...

 


Nah I don't feel bruised: they kept asking for benchmarks... I gave them some.

It seems they didn't like seeing the Phenom showing better results than the Intel chip on a majority of the benchmarks. So they used a variety of means to discredit the results:

1. Ignore them.
2. Pick and choose results they like... and ignore the ones they don't like.
3. Scream about the different sizes of memory then completely ignore the results when I personally re-run the benchmarks with less ram and post the same results.
4. Pick on any any anomalous measurements. (NO: I have no idea why the kernel compile time has changed on different runs.)
5. Claim that Phoronix is "unknown" even though it is the biggest linux suite available. (Or attempt to belittle Linux. BTW: CROCATROLL: There is a FreeBSD port.)


The bottom line: the results can be reproduced. Although I haven't seen a kernel compile time as large as the one I posted on my first stock run. Perhaps I ran something during that benchmark thinking that it wouldn't really matter. (For example: checking some of the results in a web browser. Shouldn't matter THAT much... but it seems it might have. You shouldn't do that while any benchmark is running... but hey... it happens even on Windows.) But regardless of any of that... what you have to notice is that all of the other benchmarks are basically the same scores. You know: reproduced results. That is one of the main goals of Phoronix-Test-Suite.

BUT: I don't feel "bruised" as you say. I feel entertained. They basically used every tactic that they belittle people for using if their own favorite benchmarks are not accepted. And the entertaining or humorous part is that they don't realize that they have already belittled themselves for using any of those tactics. I don't have to say or do anything to make them look desperate.

BTW: There are two "problems" with the Phoronix test suite:

1. they don't have enough results in their online database.

2. since they are still developing the process they don't show enough data for the test system. (For example: ram timings and speed, was X-Windows running during the test? Actually they might actually show this... I need to look more.)

Actually I believe most of the time people use a default Linux install with X-Windows running. But there is no way to be completely be sure of that. But it is possible that it might not matter... I haven't run the cli benchmarks without X-Windows. The benchmark suite might even not allow any to run without X-Windows. (Which would be GREAT... If it doesn't... it should. I will check.) And it might not make much of a difference anyway as long as you aren't doing anything else.
 


...but the user may already have quite a bit more RAM than they need, so adding more will NOT make the *execution* of any application any faster. About the only thing that more RAM will get you in that case is shorter application *loading* times as the required libraries and such are cached in RAM. Think of it this way- if you have 64 GB RAM and run Vista, will adding another 64 GB make it appreciably faster?

My main issue though is that when you have less memory you have a bigger chance to rely on virtual RAM which is vastly slower than physical RAM.

That is true, but such a relationship is a curve that approaches an asymptote that is defined by the workload. So above a certain point, you can keep adding RAM but you are not reducing your chances of having to swap as you already have way more RAM than you need and won't be swapping anyway.

And in a testing enviroment I expect equal setups. Meaning equivalent CPU speeds, same amount of RAM, same RAM speed and type (DDR2/DDR3), same HDD setup and so forth. That way all bottlenecks around it can be eliminated and we can focus on the CPUs actual performance.

The Phoronix Global benchmark system is not a typical lab benchmark setup where people assemble similar machines and then bench them. People go run the benchmark on their own machine and then upload their results to Phoronix's server- no matter what hardware they have. It's similar to how 3Dmark works on Windows- you run the benchmark on your machine and then compare your results to others. It's useful for identifying how one particular setup as a whole compares to another but not so much for benchmarking individual parts as the system configs differ. There is a wide variety of hardware in the results browser and maybe you can find something otherwise similar to yours to see if a certain part is faster or slower, but it's not guaranteed.
 
The benchmark he posted is interesting but is only good for those with Linux.

BTW keith, you talk about other people picking specific benchmarks to make theirs seem better yet you didn't? You probably choose this suite because it makes your CPU look better.

But still the fact was that the system Grimmy posted resulted in about the same performance as a 9850BE meaning there was something else holding the Q6600 system you posted back.
 


Or perhaps it's the only open-source benchmark suite available for Linux designed to easily download, install, run and upload results.

(There is also lbs.sourceforge.net but it is basically the same individual benchmarks with no way to easily do those things.)




It could be X-Windows. I don't know at this time because it was not a complete result set. (I need to look more. The result "shows" the X.org server... but does it verify that it is RUNNING or just look at the libraries for the version. Probably not important anyway. But I'll go check RIGHT NOW.)

BESIDES: I can accept "about the same" after having months of "AMD's best Quad can't even keep up with Intel's cheapest". People still throw that out in various forums around the internet. For example the troll 2 posts back.
 
Well the Phenom did have a 100MHz advantage and since I am not a Linux/Unix fan I can't tell if that helps the CPUs at all. But in this case it looks like they are about even which is fine with me as well. It just took Phenom a while to get there.

I understand using a easy to download and install test suite (especially in Unix/Linux since installing is not always as easy as clicking) but I wish it was a bit more widely used so that you could compare systems on an even basis.
 
i found this thread jus now.......it was very interesting with the thunderman issue.........but now its a pain..........

some apps may benifit from 4GB instead of 2GB. but most apps wont. becoz they dont need so much memory.PC will run slow in 2GB ram when u r running lot of apps together and only in this case will 4GB ram actually kik in.but if ur running just one app, surely it wont require require 4GB......
also the speed jump that u will get from going from 512MB to 1GB will not be the same as the jump from 2GB to 4GB.
but we all no amd is veri much sensitive to memory bandwith, certainly a lot compared to intel
 


NOTE: Running the first part of the benchmarks in CLI from with no X-Windows doesn't shave more than 1 or 2 seconds on all of the benchmarks up to the kernel compile. Then it is about 23 seconds faster. So a few seconds here and there... but a bit more on some. There does need to be verification that X-Windows is actually running. (It does show the X.org version even if X is not running... so that is NOT good.)

SHOCKING: I agree with you about something. (Sorry I forgot to tell you to sit down or something before saying that.) Phoronix is a pain in the butt. You have to install linux.. then download and install all of the stuff.. then run it. From the command line. (Generally INSIDE X-windows...) But a pain... especially for non-Linux users.

I've been thinking of a DVD that will boot Linux in a "live" fashion so no install is needed. You put the DVD in... reboot... type in your Phoronix name/password and maybe select the drive to use for temporary data... then you go to bed. Done. That would make things much easier.
 


It was specifically developed for Linux as there were no good wide-ranging all-in-one benchmark programs for Linux as there are for Windows. The designer, Michael Larabel, is a *nix enthusiast and runs one of the biggest *nix enthusiast site.

BTW keith, you talk about other people picking specific benchmarks to make theirs seem better yet you didn't? You probably choose this suite because it makes your CPU look better.

He probably picked it as he runs Linux and the benchmark suite is probably the best-known one for desktop Linux.

But still the fact was that the system Grimmy posted resulted in about the same performance as a 9850BE meaning there was something else holding the Q6600 system you posted back.

Perhaps there was, perhaps there wasn't. We don't know what else might have been running on the person's machine. But here is a list of Intel quad vs. AMD quad tests from the same site:

Q6700@stock, 2 GB RAM, Ubuntu 8.04 amd64: 674.92 sec
Q6600@stock, 4 GB RAM, Ubuntu 8.04 amd64: 572.86 sec
Phenom 9750 @stock, 4 GB RAM, Ubuntu 8.04 amd64: 651.74 sec
Phenom 9750@stock, 4 GB RAM, Ubuntu 8.04 amd64: 652.25 sec
Phenom 9750@stock, 4 GB RAM, Ubuntu 8.04 amd64: 666.39 sec
Q9550@stock, 1 GB RAM, Ubuntu 8.04 i386: 766.87 sec
Q6600@stock, 2 GB RAM, Ubuntu 8.04 i386: 848.80 sec
Phenom 9850BE@stock, 8 GB RAM, Gentoo amd64: 528.85 sec
Q6600@stock, 4 GB RAM, Ubuntu 8.04 amd64: 585.73 sec
Q9450@3.33 GHz, 8 GB RAM, Ubuntu 8.04 amd64: 1044.55 sec (bench version = 0.7)
Q9450@3.20 GHz, 4 GB RAM, Gentoo amd64: 383.66 sec (bench version = 0.5)
Phenom 9850BE@2.70 GHz, 2 GB RAM, Debian Lenny/Sid amd64: 615.63 sec (bench = v0.5)
Q6600@stock, 2 GB RAM, Ubuntu 8.04 amd64: 585.73 sec (bench version = 0.5)
Phenom 9850BE@2.75 GHz, 2 GB RAM, Debian Lenny/Sid amd64: 630.79 sec (bench = v0.5)
Phenom 9850BE@2.70 GHz, 2 GB RAM, Ubuntu 8.04 i386: 698.05 sec (bench = v0.5)
Phenom 9500@stock, 2 GB RAM, Ubuntu 8.04 amd64: 968.15 sec (bench = v0.5)
Q6600@stock, 8 GB RAM, Mandriva 2008.0 amd64: 541.10 sec (bench = v0.4)
Phenom 9850BE@2.80 GHz, 2 GB RAM, Debian Lenny/Sid amd64: 429.85 sec (bench = v0.4)

Here's what I gather:

1. All processors are slower under i386 Linux than they are under amd64 Linux.
2. The Q6600 at stock on an amd64 OS compiles the kernel in about 580 seconds as there are several different Q6600s in that range on both v1.0 and v0.x benchmark versions.
3. The stock Phenom 9750 compiles a kernel on an amd64 OS in about 670 seconds.
4. The Phenom 9850BE at 2.70 GHz with the benchmark version of 0.5 compiled a kernel in about 620 seconds, compared to about that same 580 seconds for the Q6600.
5. We only have one data point for the Phenom 9850BE with the newest benchmark version and it doesn't fit very closely with the results from the older benchmark version.

So overall, the Q6600 is > Phenom 9750 but we really can't compare to the Phenom 9850BE at stock due to a lack of data points of the Phenom 9850BE at stock and using the same benchmark version and OS- Gentoo can have some big differences between computers.
 
Hey, keithlm! Long time no see! : p Glad to see your posts without having to "Show comments: All" - although the guys still have been a little "picky" about your comments.

Interesting benchmark. It would be nice to be able to run it in a "live fashion". Just got my Ubuntu 8.04 (both 32-bit and 64-bit) CDs, so, perhaps I'll download Phoronix.
 


OH YEAH... DO IT! They need more Phoronix results. Especially if you are going to have a Q9550. Stock and overclocked.
 


Sure! : D But, anyway, I noticed yours is such a fine rig. How is it going? BTW, have you tried some 64-bit OS? Phenom is supposed to perform a little better in a 64-bit enviroment, but that's just the rumors I heard. Well, I got excited about the new SB750 - especially since Anand had such an "easy time" increasing the NB and HT speeds -, however, there weren't any published benchmarks with the improved frequencies. I would like to compare the same Phenom rig just with different NB/HT frequencies to see if that might be a bottleneck. Unfortunately, I don't have one. Guess I'll try something similar with the Q9550, like increasing the FSB to 450 while decreasing the multi, to see if the FSB is really limiting the performance of C2Qs.

Do you think you could run some benchmarks in the conditions I stated - even without the SB750 -, keithlm (CPU stock frequency; NB/HT overclocked)? I'm really looking forward to see the comparison!

Edit: just something like H.264 encoding!
 


What interests me is that RAM size seems to have a significant effect-- more so than the minor frequency delta between Q6600 and Q6700. Unfortunately, we only have the single Q6700 and it might be an outlier.
 


The Gigabyte DS5 board is not completely stable with a 9850. Even at stock NB speeds. Attempting to increase the NB requires a CMOS clear to recover.
 


Unfortunately I doubt there will ever be a sufficiently large enough collection of data on all of the processors that any real conclusions can be made. As a result the debate will likely rage on until the new processors from AMD and Intel come out and then nobody will care about these results anymore and a new argument will start.
 


well said. +1
 


I think the old Athlon 64 I have does perform better under a 64 bit OS, I just upgraded the kids computer to Vista 64 from plain old XP and while I noticed no performance increase neither did I notice a performance drop (this tells me that while Vista is a resource hog being a 64 bit version helped ease the pain a bit)

I should mention the specs 2GB's DDR Athlon 64 3500+ socket 939 with a Radion X1600 Pro 512MB's

The reason for the upgrade ? some spyware posing as an antivirus.... (even I could not kill it completely and I have up to date version of almost every anti virus tool made... even all of them renamed so they can run on infected systems.)

Anyway I think AMD does have an advantage over Intel on 64 bit from my experience. Until I get my Q6600 so I can get on the Core2Duo bandwagon and see if that makes a huge difference I am still running a Pentium D OCed to 4Ghz for my machine.

I know Core is faster but how much faster then 4Ghz ? lol Is it OMGWTFBBQ faster ?

Edit: By the way the one thing about AMD that I have a complaint about... I would like to buy a X2 939 socket CPU... I think AMD dumped that socket a little too soon and may have hurt them selves in the process.
 
^Its hard to say what you said since you haven't been able to compare. And to say the truth you wont be able to unless you have at least a Phenom 9750BE. The Q6600 will easily beat out your older system even in 64bit.

And you are very right. AMD made a mistake by dumping S939 so fast and screwing over a lot of people like yourself in the process. I don't see how they couldn't just adapt S939 to DDR 2 and why the extra pin was that neccisary. But what can ya do.