INTEL - Synchronous is not important

[DAPUNISHER]

Originally posted by: stevejst

Ad Hominen attacks won't win you any credibility in these forums troll

Apparently it holds you in a cage though I guess you did not get your banana today. Cool down gorilla.

I'm just havin' some fun with you Steve, it's just the intardnet, ya know? As the saying goes, Arguing on the intardnet is like running in the special Olympics, even if you win you're still a retard * no offense to those with handicapped relatives intended or meant*

 

[stevejst]

youre type of attitude is why many ppl no longer bother to post information here.

No. I posted 800 or so times. You and the gorilla posted combined 16,000+.
If somebody is turned away from this place and by somebody, that is surely because of the intellegentsya like two of you.

You do not know anything but to sit here 24/7 and talking high to each other.

@dapunisher:
Cool down, again. You are not as tough and as provoking as you imagine yourself.

 

[THUGSROOK]

thats right.

and we know what you wanna know.

so if you wanna know it too ~ be nice

 

[DAPUNISHER]

Didn't read my last post eh Steve?

 

[stevejst]

thats right.

and we know what you wanna know.

so if you wanna know it too ~ be nice

No, actually you don't. You only know what you have in your eclectic bag from other "smarts" and OC web sites.
That is exactly how you learned what to uncheck and check in Sandra. I asked you simply what disabling of SSE has to do with buffering but of course you have no idea because that is how much you know about programming and assembly in the first place.
Who knows, if you visit back that site where you discovered the image you posted you might even learn that.

I doubt you have any technical or hard science knowledge whatsoever. If you know what is important to know you would not be sitting here lecturing for 8,000+ posts. You would be able to sell that for money and buy better PC than the one you are so proud of. Got it?

Dapunisher, take a hike, I am not interested in your posts, get lost.

 

[DAPUNISHER]

I told you you could stop dancing fool! yet you continue? Oh well, just another troll who will soon be banned for being obtuse

 

[stevejst]

 

[Mikki]

Originally posted by: stevejst

thats right.

and we know what you wanna know.

so if you wanna know it too ~ be nice

No, actually you don't. You only know what you have in your eclectic bag from other "smarts" and OC web sites.
That is exactly how you learned what to uncheck and check in Sandra. I asked you simply what disabling of SSE has to do with buffering but of course you have no idea because that is how much you know about programming and assembly in the first place.
I doubt you have any technical or hard science knowledge whatsoever. If you know what is important to know you would not be sitting here lecturing for 8,000+ posts. You would be able to sell that for money and buy better PC than the one you are so proud of. Got it?

Dapunisher, take a hike, I am not interested in your posts, get lost.

Wow. Where do these people come from?!! "science knowledge"? Anyone care to explain what that has to do with anything? So all the hours of testing and tweaking don't mean anything because we don't have "hard science knowledge"? That's it!! I give up! Guess I'll have to take on a less "technical" hobby....

What a maroon... I'm enjoying the dance tho....

 

[stevejst]

Wow. Where do these people come from?!! "science knowledge"? Anyone care to explain what that has to do with anything?

Explanation? See if you are not a programmer, you can hardly understand what SSE and SSE2 are. If you do not understand that, you would have no idea how important or unimportant that is in "everyday" computing or code optimization. If you do not understand that how would you know how to tweak or testbench your setup? You can only tap in the dark or believe what somebody else tells you, right? That is what these two guys are doing, it is not hard to notice. They think they are very smart because they read somewhere to uncheck number of options in one Windows application. They themselves have no clue why and tomorrow they might change these options if somebody else tells them something else.
And that is only one "small" thing.

I said once and why not repeat - I see no reason why to uncheck any optimization since the reality of modern software and PC is the optimization. You will not run Windows applications unoptimized, as simple as that.
The "thugsrock" guy posted "unbuffered" scores of his memory the way he read somewhere "unbuffered" suppose to be so he thinks he knows a secret and the rest of us don't. That is foolish because the reality is that he himself has no idea about that in the first place.

These guys are debating code optimization/unoptimization and they themselves have no clue about that. And why they don't have a clue? Because they do not know anything about code optimization, assembly, or even chip architecture they are so ready to lecture about. That is the hard science and that is the point.

 

[Slappy00]

I took two years of computer science too to get my minor and I consider myself a decent C++/Java programmer (assembly is for the nutcases though). I love it how ppl will flex their alleged computer science knowledge, when they like many other computer scientists: cant live without some sort of reference book and hated every minute of assembly.

But I degress 1:1 is the best setup #1 because my left nard hanges ever-so slightly lower than the right one, and when it talks to me it says that 1:1 is better and Old Milwakeee Light is wayy better than Keystone....

 

[stevejst]

I took two years of computer science too to get my minor and I consider myself a decent C++/Java programmer (assembly is for the nutcases though). I love it how ppl will flex their alleged computer science knowledge, when they like many other computer scientists: cant live without some sort of reference book and hated every minute of assembly.

I am actually teaching both C++ and Java, and I didn't do any assembly for ages. I don't consider myself computer scientist though. Most of my students get A and B but I wouldn't vouch for most of them to be computer scientists either.

 

[Mikki]

So you have to be a programmer to know what you're talking about when it comes to overclocking? Are you serious? Have you actually tested all the benchmark settings like we have or are you just relying on your "superior hard technical knowledge" when you post this stuff?? :Q

 

[stevejst]

So you have to be a programmer to know what you're talking about when it comes to overclocking? Are you serious?

Apparently not. Most of overclockers are not programmers, I guess.
But if you want to debate synthetic benchmarks I think you ought to know what they are, wouldn't you say?

 

[Technonut]

But if you want to debate synthetic benchmarks I think you ought to know what they are, wouldn't you say?

Perhaps you could enlighten us regarding the pros and cons of running buffered/unbuffered Sandra benches and how each relate to SSE(2) instructions, data pre-fetching, and caching.....

 

[TerryMathews]

Well, I would imagine that most CS grads would know more than a little bit about RAM and how it works as CS has to take some CEG courses. At least, at decent colleges.

Anyone who truly understands how DRAM works will immediately see how asynchronous clocks introduce latency. Latency, obviously, is something that SANDRA doesn't benchmark.

 

[stevejst]

Perhaps you could enlighten us regarding the pros and cons of running buffered/unbuffered Sandra benches and how each relate to SSE(2) instructions, data pre-fetching, and caching.....

Enlightment is something you'd have to pay for.

 

[stevejst]

Anyone who truly understands how DRAM works will immediately see how asynchronous clocks introduce latency. Latency, obviously, is something that SANDRA doesn't benchmark.

It does. But in fact I posted two other tests here and they are in accordance with the claim. I said, synchronicity is worth anywhere between 2 to 6 MHz of FSB, depending on the test. So there is a loss but a loss which is amptly compensated by aggressive timings you can run at the lower clock for memory.
You can easily repeat my tests yourself, granted you have the processor and similar type of memory.

Here is another (4th) test, Using AIDA32, which is inferior to Sandra.

210:210 (synchronuous) read 4979, write 1750

236:190 (3 MHz up asynchronuous) read 4880, write 1750

Try yourself if you do not believe my numbers.

 

[TerryMathews]

Latency can have ramifications beyond measurable bandwidth. Which was my original point. For example, you might end up with a system that gets a great avg FPS but still stutters.

 

[stevejst]

Latency can have ramifications beyond measurable bandwidth.

That is possible and you might be right but I do not see how to measure that.

The whole point is if you have a system that can have 280+ FSB and memory that cannot go anywhere beyond 220-230, there is no point in sticking to sync clock. Because that will keep you at 220:220. If you use 5:4, you'll be able to run 275:220 and that will outgun 220:220 any way you take it. 260:208 will better already and with 2.4 GHz and 2.6 GHz processor you can run it. Beside on 208 you'd probably use more aggressive timings.

 

[WarCon]

Just curious about modern C++, are the SSE (2) instructions separate or does the compiler just implement them?

 

[stevejst]

Depends on the compiler and IDE. The ones I am using (gcc and g++ on Linux) do not, as far as I know.

 

[WarCon]

Been over 15 years since I programmed (love/hate thing with programming so I quit). Actually liked Pascal over C+, but guess which one won....

 

[TerryMathews]

Originally posted by: stevejst

Latency can have ramifications beyond measurable bandwidth.

That is possible and you might be right but I do not see how to measure that.

It's not measurable. I said that in my first post. But it's obvious if you know where to look and what to look for. Also has the tendency (especially in gaming) to strike at the worst possible times IE when there are a lot of enemies in close range. Very similar to gaming over old dial-up connections. You'll be fine until you run into someone and then BOOM lag-city.

Actually, now that I think about it, you might (emphasis on might) be able to measure this IF you have a frame counter that's very small and lies outside the game's thread. It'd have to measure the frame count from the video driver and not the game engine. Then, stuttering should be indicated by times where the instantaneous FPS is at or about 0.

 

[infinite012]

How much longer can these nerds last!?? haha

Anyways, I'm agreeing with Jeff only because he's helped me so many times. I'm too lazy to read why you all are arguing...all this babble (4 pages!?) in 2 days..

EDIT: haha ok I read the title. Synchronous is important because steve misspelled it

 

[stevejst]

Synchronous is important because steve misspelled it

Uh, got me on that one. That is mathematics induced spelling (aka continuous).