Intel vs. AMD

PacMan777,

First off, thanks Sam for that quote as you are right. I didn't say the 6400 was better only that it overclocks better. Second, I've heard lots of claims for 3.5 to 3.6GHz on air, but I've yet to see benches to support it. Sam and I both know a number of very knowlegable people that are presently tweaking thier 6600s and no one is close to 3.6 yet.

Here's a quote from Anandtech on the subject!

Now I'll be the first to admit that the 6400 was a better "bang for the Buck" CPU last week than it is today because the gap in price between it and the 6600 has dropped to below $100. I'm of the mindset that 300MHz isn't worth the price difference to me. I agree with Sam there! I'de put that $96 difference toward another 2 sticks of 512MB memory.

On the $7,000 price tag, I meant to say that his systems compare to and often exceed the performance of a 5 to 7 thousand dollar custom Alienware rig. I had to go back an read what was said so my error! Sorry!

What we really need is to see Tomshardware or Anandtech do an updated comparisun with the better motherboards and memory available and see what thier techs can get out of them. When I do build the 6400, I will be posting benches so if I fall on my face you guys will be the first to know. I'm expecting 3.2GHz, which is a 50% OC! I'm pretty fair at overclocking Intels as I've gotten 3.94 on air with a E630 P4 3.0/800 Prescott and 4.1 on air out of my D-940, both stable. I know I can do better with the D-940 but it will take a better MB with better voltage regulation than the P5P800 SE I'm using now. Great MB when I bought it, but technology has passed it by! That's why I run it at 3.71GHz folding at 95% CPU usage. Folding is one of the best tests for stability. Better than Prime 95 or even the new kid on the block, OCCT! I've run both for 24 hours without any errors!

Stay Tuned,
theone

 

Your math is off a bit. Actually if those figures are correct, the 6400 is OCed by 60% going from 2.13GHz to 3.4GHz. The 2.4 to 3.6 is an easier calculation at an even 50%. But getting to the actual differences at those percentages, note that the E6400 (2.13GHz) improves by 1.27GHz and the E6600 (2.4GHz) improves by 1.2GHz. That's 70MHz that the 6400 OCs better than the 6600 while the 6600 OC is a total of 200MHz greater. Assuming the extreme OCs are correct which most of us won't be using, I'll still take the 0.2GHz and 2MB cache difference anyday for an extra $100 compared to any of the other noted model differences.

 

You would, yes, because you can afford it, but for people with less to spare, I think 2GB RAM would be more important than a bit of extra cache and mhz. It gets me that THG havent reviewed a 6300 (to my knowledge) and done the extreme overclocking test on it. They don't even include the 6300 in their supposedly comprehensive CPU charts, I have to use the 6400 and guesstimate.

As far as I can see my maths isn't off, I just used a different example, choosing 3.3Ghz rather than 3.4Ghz as the overclocked value.

 

Sammorriss
Excuse me on the 3.3 and 3.4. With both you and theonejrs, I became a bit confused. The difference between 50 and 55% would be even more negligible. The end numbers I used, which are pushing those CPUs, shows only a negligible difference in OC ability and the E6600 as a clear winner. To each his own and we'll budget the money where we think most important. But for the sake of comparing CPUs, it's obvious where the bang for the bucks lies. By all means, for those who are on a strict budget, get a 6300, use the cheapest RAM possible, along with a cheap board and buy the cheapest case and psu available. Do the budget build up right. Out of curiosity, how much did your CPU cost?

Theonejrs

That's a questionable example for our discussion. They're comparing OCed processors to stock models without saying where those stock models can go. The review shows the larger cache is useful in memory intensive applications which agrees with what I said. The only thing of note is that the author concludes "... the 2MB Core 2 chips preform respectably regardless of the application being tested". No one appears to disagree with that. So, what's the purpose of the quote?

 

That sounds familiar. LOL Just go back to the closed AMD-Intel thread. I browsed that one as well as this one. The everyday working PCs are the majority of sales, not the performance models enthusiast builders prefer.

To your list of vendors you should add Dell stores and their online sales along with Gateway. They make up a lot of sales for the average consumer, plus they handle a lot of group sales such as businesses and government offices.

Your words are true and you should add most users won't be overclocking or fully using the multitasking capabilities of their PC. For many users a good socket 478 P4 would be more than enough. I'm surprised we haven't seen the new single core that was promised.

As for Intel and the pricing, I wonder. Looking back, everyone was paying some stiff prices on AMD processors that don't hold a light to Conroes that are selling for half the old prices and near what those high priced lower performing AMD processors cost now. A good example would be the 4800 X2 and the E6600. Last year the 4800 sold in the "sticker shock" category and it still sells for over $260. Newegg has an OEM for $250. I used one not too long ago that was still $300. Apparently the pricing is continuing to drop on the old tech processors. Meaning even less profit for AMD.

I don't know what the market will bear, but Intel has yet to test it. The Conroes are at giveaway prices compared to what the AMDs were not long ago.

 

I had just seen on Intel's website two different versions of a more advanced Intel Core 2 Duo, the Intel Core 2 Extreme, this one is sold in dual core and QUAD-CORE(has anyone found a computer with that processor, the quad-core Core 2 Extreme?). Should I save even more for one of those or will the Core 2 Duo E6600/E6800 be enough for now(I want to keep my computer for a long time, as the one I have, a cheap 800MHZ Pentium III, is outdated, bought 7 years ago, in 2000)?

 

There's a huge price gap between the E6600 and the X6800. The X denotes Extreme. There is an E6700 for about $200 more, but the X6800 goes for about $950 US. The quad core QX6700 is also an Extreme selling for about $1000. I've not noticed an E6800, at least not yet. Had you looked up the prices, that should have been a deterrent.

Tests have shown the quad core holds no significant superiority over the X6800. Check Tomshardware and Anandtech for reviews. There are no apps that give quad core an advantage currently in the real world, so the jury is still out. If I had the bucks to spend and a need for the Extreme processor, my choice would go with the X6800 over the QX6700. But I wouldn't go there. I'd build a nice E6600 or splurge for an E6700 at about half the cost of an Extreme.

After all that, my suggestion is build a Core 2 Duo that is reasonable within your budget. Whether an economy build with a lower model Core 2 Duo or a more performance oriented starting with the E6600 you can't beat the present Intels. Even the E6300 will be mind blowing in comparison to your current P3.

 

PacMan777

The purpose of the quote was to be sure everyone knew and understood that we were talking about OC'd compared to stock. I could have quoted just the line about the cache, but I try to be fair and not try to skew the results! You will note that's how Sandra shows the results, your overclock vs stock speeds of various processors. True, you can overclock the higher end processors to better performance too. It's just that there are too many variables to just say you can OC a 6600 to 3.6. Choice of MB, memory, Power supply and so on, not to mention the skills of the builder.

As far as percentages go, I just take the processor speed for the 6400 at 2.13, multiply that times 50% and add the result to the base to get 3.195. It's the same way the MB manufacturers do it. If I take my A64 at 2.40GHz, multiply that times 15% and add the result to the base 2.40 I get 2.76GHz. If I set the automatic to 15% in the bios set-up I get 2760MHz which is 2.76GHz.

I would like to point out something to newbies reading this thread. None of the current dual-cores are slugs, except maybe to the hard core enthusiast. Any dual core will run rings around the fastest single core. If you decide to build a computer buy the best you can afford. This applies to the CPU only!!! Your performance will be directly geared to the quality that you support that CPU with. You can't buy cheap memory, motherboard and power supply and expect good numbers, no matter how good the processor is. You can build a better PC than you can buy from Dell, HP, Compaq, Gateway and e-Machines but you have to use quality components for the best result as you can't make a race horse out of a mule!

What's that footnote you always see? "Results may Vary"! Actually there are many more variables to be considered. For instance, with CPUs there's binning, stepping and electrical tolerence to consider. A guy that gets the better binning or better stepping will have the faster computer using the same MB. Differences in the allowed tolerences will also make a difference as no two electronic components are exactly equal. That's precisely why NASA was precision grinding resistors with high speed fine diamond dental drills back in the 70s and 80s to get the best spec possible for thier needs. It all boils down to a certain amount of luck. If you wind up with the best binning, best stepping and best tolerences in all of the components, you wind up with the best machine. At the other end If you have something where all the tolerences are within spec, but on the low side, then that machine will not be as good!

Happy Computering,
theone

 

The Gigabyte GA-965P-S3 and the other parts are here. I got it Friday. But a friend came over and we've been hanging out. So I'll build monday night and tell how it runs.

 

the best of luck with the new build Estuansis

 

Estuansis,
Good luck and enjoy your new toy!

theone

 

I'll be interested in what you do with the RAM.

Theonejrs
I'll not qoute you. That reply was a bit long. Rule 1: you don't compare OCed models to stock ones unless as a guage or improvement for the OCed CPU, much as the SiSandra benches; Not As A Comparison Between The Two. If you're going to say things like an E6400 outbenches the E6600, give a link or show the benches like I asked with both being OCed on the same test boards. You may have put some PCs together, but building for performance and doing comparisons is a different arena. As the end figures I provided show, the difference is insignificant and the E6600 has a noticable superiority over the E6400.

Talk budget, but $100 isn't that big when you're talking about a PC over $1000 US. So, as most people agree, the E6600 is the best overall bang for the bucks. We all know that the lower end Conroes are good, and much better than what was past. My suggestion to Ace_2 just that, but thanks for restating it ... again.

Thanks also for adding all the little variables that affect performance. That does cloud the point though. What was being stated was that test results should be derived using the same test boards and keeping the hardware the same except for a particular piece of hardware being compared. Variables down to the silicon used make a difference, but those are out of the control of the enthusiast. What we can do is compare the performance between 2 processors keeping the test variables as close as possible and not skewing with the use of different RAM or a different board. The only thing we can do with binning is try to find a good one, such as some of the Opteron 170(s) not long ago. Throwing in the trivia only muddles what was being said. When testing, keep the variables limited so the item being tested will be giving the results, not a group where each component can add to a change.

Seems we mostly agree on the processors, if not on the budget and some of your questionable test methods.

 

Theonejrs
You were talking about benching and doing the torture tests. What do you do about "SpeedStep" and "Cool 'n Quiet"?

 

PacMan777,

While we do agree on some things, I can't agree with rule #1. The sole purpose of the Sandra benchmarks is to compare your overclock to known stock benchmarks so you can see what your overclock matches up to in terms of stock processors. While it would be nice to own all the high end Conroes and AMDs, as it would be the only real performance test head to head, it wouldn't be practical for most people to do. Back before I got sick I could well afford to do it. I'm no longer able to afford it.

I'm not complaining as I've had a wonderful and very adventurous life. I chose to liquidate everything and take care of my children's future now while I can enjoy it with them without them having to fight the Government for years in court after I'm gone, for what should be rightfully be thiers to begin with. I hadn't counted on being fourced into early retirement so things changed drasticly for me since I had to retire last May for health reasons. After years of 5 and 6 figure income I,ve had to learn to count my pennies to get the most out of my money.

As far as my saying that the 6400 outbenches the 6600, I never said that! I said it overclocks better, meaning percentage wise as Sam understood. When Ace_2 asked the question, he asked about low end AMD and Conroe processors. I only suggested the 6400 because of the small price difference between it and the 6300 that would yield significant gains for him performance wise without having to spend much more money. At the time the 6600 was $165 more and probably out of the question for his indicated budget. You could make your case for the X6800 on the same basis. After all it's only $958! If we all bought one there would be no need for this thread!

I have one more thing to say an I'll drop this discussion.

When I do a build, I like to take my time and be very through in my personal test methods. I use Sandra, Cpuz and OCCT and provide benchmarks. I usually run Prime 95 and now OCCT for 24 hours and then I set it up for folding. I'm only satisfied if it passes all of these tests and folds without errors. I don't bench race as I think it's stupid and totally impractical in the real computing world. Since I first joined this forum this thread has always been about making lesser processors run more like higher end ones. Everyone understands that you can overclock the higher end ones as well to even better benchmarks. All that takes is more money! We even have several folks here that have more than enough income to be able to afford any processor on the market. Yet most of them enjoy taking a lesser processor and seeing what they can do to improve it's performance. It's a total learning and sharing experience that we all enjoy doing. This thread is made up of a great bunch of folks from all over the world who enjoy the experience and thrill of seeing thier computers become all that they can be. I have a friend in another Forum who is folding with 5 different computers he built, at the moment. None have top of the like CPUs. A couple of Opterons, a Smithfield, a Pressler and a Conroe. He could well afford all X6800s, but there's no, challenge for him in that. That's what this thread is all about!

Opps, you asked another question!

I usually turn them off in the setup if the bios allows it. I've only come across one that you couldn't and that was on a Panasonic Toughbook CF-48X with a mobile P4 that I recently repaired.

Happy Computering,
theone

 

as much as love this banter i think somone needs to step in before the mods do and this thread gets closed again.

this is the bitchy intel sucks no amd blows thread, also known as intel vs amd.

you are wandering into the dark art of overclocking, which as you both know has its own place.

come on guys we all like this thread but going off topic was why it was closed last time. if you want to talk ocing go into that thread. if you you want to box for the green or the blue corner, ding ding.

 

Marsey 99
I don't see where you get the thread is going wrong and there is a need for a mod. If there is never disagreement, then the thread is not likely to be very useful, except for telling others how to insert their RAM. Neither theonejrs nor I are attacking each other, just discussing opinions on processors and testing methods. I regret you got the uncomfortable feeling there was a flamer pending. I'd leave the thread first as Kivory666 did. It's not worth the hassle.

I don't see overclocking as a dark art. It is one reason why enthusiasts like all those boards that include the ability. The reason the manufacturers lock their BIOS(s) is to keep novices from damaging systems and costing them in warranty claims. Interesting is the fact that some board manufacturers will honor the warranty when it's obvious the user fried the solder on the circuits by turning the power too high. Any thread dealing with processors and platforms is going to get a question or 2 about OC(ing).

As far as blue and green go. I don't think that is an issue. History proves the facts on that one. I don't have a favorite as I've built with both. Intel was the leader for the longest with their PC processors. AMD came along and got into the market and played second fiddle for a long time. Then they came up with Hypertransport. At the time Intel was using Netburst and switching from the Northwood architecture to the Prescotts as their main processor. We all know how Intel suffered in comparison to AMD following that. This past summer Intel released the Core 2 Duo (Conroe). Magazines such as PC World gave the Intel Core 2 Duo architecture credit as being among the top innovations of the year. That's the short history of the situation with no bias toward either AMD or Intel.

If we're discussing AMD and Intel how can we be going off topic?

Edited to add content.

 

PacMan 777

The only Core 2 Duo I was able to come across is the cheaper 1.86GHZ E6300. Here's the computer I found with that: http://www.futureshop.ca/catalog/proddetail.asp?logon=&langid=EN&sku_id=0665000FS10080558&catid=
You could also PM me with your thoughts on all the other aspects. After all, this is a PROCESSOR discussion thread.

 

Ace_2
I suspect discussing a platform falls within the thread, since most other things get included. I just noticed in a January review in PC World on PCs there was a Micro Express Microflex 668 with an E6600 and it comes with a 17" monitor for $999. It was listed as #1 best buy in PCs for the month.

 

theonejrs
Had you not misunderstood what I said about Sandra benches, we would be in closer agreement. We pretty much said the same thing only in a different way. Sandra benches are just a measure of how a system we're testing stacks up against the stock systems in the Sandra database. Plus, we get readings on the tested system that we can compare to other readings we've gotten with other settings. Most helpful when OCing and tweaking a system. When talking about overclocking Sandra still shows how the tested PC stacks up against stock, but the results are skewed as far as comparing OCed against stock (apples and oranges). There is no data in the Sandra database for OC(ed) systems (just stock). You'd have to go to test systems and run individual benches for comparing OC(ed) processors because that can't come from the Sandra database. Unfortunately, I don't have the funds to test a full line of AMD and Intel processors.

I also regret your misunderstanding where I was going with Rule:1. Maybe I should have stated it differently. It's not a "rule" I'm trying to set up for anyone. What I meant is that anytime there is a scientific or quasi-scientific attempt at comparison of items, such as processors, the variables should be isolated as much as possible. That's standard analytical procedure. That's why legitimate test centers use test boards set up the same and also strive to have system drivers as close to the same as possible. That's sometimes difficult between AMD and Intel, but poses no problem when comparing different models from the same line, such as various Core 2 Duo processors.

Outbenches, overclocks; take your choice. The ability is very slight between the E6400 and the E6600, but the significance is in the overall performance superiority of the E6600 over the E6400 for about $100 difference. I gave the end figures which mean more than just swapping percentages. We can leave the rest to enthusiasts' budgets. In the real world and even according to Intel, as you mentioned earlier, the E6600 is currently the best CPU for the money. Not saying everyone wants to pay $318 for a processor. I suspect many of us here have paid $250 or more for a processor, but that's budgeting, not the quality of the processor.

I would have said set the BIOS to manual and then turn off the "speedstep" or "Cool 'n Quiet" (depending on whether Intel or AMD respectively) for the benefit of any novices reading the thread. But most of us know what you mean. If not done, the test is invalid because the system will never be fully loaded. Just including that for the novices out there. I wasn't aware you'd gotten into AMDs and started learning their BIOS setup. Building and working with systems gives more experience than just comparing benches.

Unfortunately there are those here who appear to want to protect their turf or become argumentative over little to nothing. I know I don't know it all and came here to share what I know and hopefully to learn a few things. So far what I've observed is the best you've had on here for a while get driven from the thread by poor behavior. It doesn't take a rude outburst like the one that drove away Kivory666. I won't tolerate the behavior either. I'd rather just stop posting to the forum.

 

Yes, well said, this thread's been plagued with that time and time again. You've conducted yourself well in the discussion pacman.