The Official OC (OverClocking) Thread!

Sam,
I understand and fully agree with you, but all I'm saying is that AMD demonstrated a successful working model of 45nm Quad core architecture, before Intel did! Even though it was more than likely a "one of", they were still the first to accomplish it. It was a milestone in CPU technology, nothing more! The fact that the public didn't get them until 14 months later doesn't change the accomplishment.

Best Regards,
Russ

 

To be honest, I wouldn't be surprised if I heard Intel had made a 45nm chip but didn't display it because they weren't happy with it...

 

Sam,
Intel was having it's own problems at the time with 45nm. That was part of the delay they had in bringing them out! They couldn't get it to work! That's why they went to the Metal gate Transistors. I don't know if it was the C2 design over the Monolithic design of the AMD Quad, but I do know those Transistors solved the problem for Intel. AMD had their own problems as well. Poor clock speeds and the TLB error among other things. Both were late to market.

Russ

 

Indeed, but what's really important is who got to market first, and Intel beat AMD by a huge leap. That wasn't really necessary in terms of performance, only when the Phenom II X4 940 came out did Intel have competition against any of their 45nm products, but it had a dramatic effect on power consumption.

 

hmm it appears people have been forgetting to take note of this particular message at the top of the page 'NO Fanboy comments needed'.

Does it really matter who beat who we all bought whichever chip for our own reasons. If you want to argue aimlessly go elsewhere.

Peace out.

 

Not really, it isn't a case of fanboyism comments (at least I don't think so) - more an explanation of why the market panned out like it did.

 

Its needless arguing no matter which way you look at it.

 

I wasn't trying to argue... :S

 

On a lighter note rather nastily and from an outside looking in perspective on Sims 3 it is impossible to achieve faster speeds via OCing you end up with one of the following predicaments:
The computer breaks
You are told there is no effect
You are told 'its not running any faster but the graphics somehow look better'
or finally 'its not running any faster but it certainly looks more stable'

How unfair is that bearing in mind this a PC exclusive...

 

So wait, you have the option to overclock PCs in The Sims? Whatever next (Oh, and as a developer's standpoint that's probably what they were paid to put in the game)

 

Interesting (yawns to self) debate, but no one has really provided any links to verify just who was the first to design a 45nm CPU. If I had to choose sides I would have to go with Intel since there were rumours that go back at least a year before they released their first which was in turn released more than a year before AMD released on. Any company can spin a yarn about having a working processor but until they actually release one it is all vapor, smoke and mirrors. AMD announced that they were going to release one in 2007 but it was a year later before it happened.


AMD indeed was the first to release a true monolithic Quad and I expect more first from AMD because they are true innovators. Without them many of us would still be running some variant of the P4.

AMD's first 45NM chip was released in November of 2008 or about 8 months ago. By then I had already replaced my 45NM E8400 with an E9650 and Q9550 EO stepping. In the end however none of it matters. All that I want to know is which overclocks the best and provides the most performance. I don't overclock for sport I do it for practical reasons such reencoding Blu-Ray to fit on a standard single or dual layer disc.

http://www.custompc.co.uk/news/605178/amd-launches-first-45nm-cpus.html

http://www.tomshardware.com/news/intel-announces-45nm-sram,2130.html

 

Overclocking is an interesting debate, because recently I would say in almost all sectors of the market up to £180, AMD vs Intel is a tough call. The AMD performers are slightly cheaper, but use more power and the big one is that they don't overclock as well. I've seen some high numbers from the recent chips, 3.7-3.9Ghz sometimes even 4.0 or 4.1, but taking into consideration most of these chips are 3Ghz+ to start with, the fact that 2.66-2.8Ghz Intel rivals can reach these numbers as well skews the balance somewhat. It's also a matter of manageability, the AMDs often take big voltage hikes to reach these figures. The Intels can often do it without any settings that would compromise long-term reliability. As a stock CPU, the X2 7750 and 7850 are better value than the E5200 and E5300 but barely overclock by more than 15%, whereas 60% isn't especially hard with the E5 series. Again, the Phenom II quad cores, very competitive CPUs on their own against the Q9400 and Q9550 (though the latter is now cheaper over here than the 955), but overclocks are rarely much over 25%, whereas the Core 2s will happily manage about 40 in a lot of systems, provided they're well designed.

 

As I stated overclocking is only part of the picture, the other is performance. If I decide to build today I might well go for an AMD but already having a highly clocked Q9650 and Q9550 it would be throwing money away. As far as I know the Q9650 and Q9550 maximized are still faster than the best maximized AMD. Until there are 6 core units there is really no reason for me to upgrade

 

The Q9550 and Phenom II X4 955 are actually impressive rivals. As always with AMD vs Intel and video encoding it depends on the method you use but in some tests the 955 gives the Q9650 a fair fight. That said, if you've got the Q9550 or Q9650 any further than about 3.6Ghz, you'll flatten the AMD regardless, and 3.6Ghz isn't even difficult to achieve with a C1 stepping Q9550, let alone an E0 9550 or 9650.

 

Well it's a given that Intel chips are still on top ATM. Core 2 technology has been around 2 years longer and they've got it refined and have maxed out the current architecture or very close. That's why they released i7, which isn't all that practical to build with anyway. The same will happen with Phenom II soon too. Maybe they'll match Core 2, or maybe they'll surpass it. But that's all that's left to do with the architecture, refinements. Last I heard, both companies had difficulties doing 45nm to begin with. So anything smaller will take a while. Mind you, Phenom II is a lot newer so they have more room to improve.

Also, if you're going for absolute speed, you'll be buying Intel either way. But when you get into the more mainstream segment, competition is fairly close. That's why I've chosen to keep quads from both sides, and both are damn good chips. No matter if my Phenom II has a higher stock clock. It still hit the same 3.7GHz my Q6600 did and both can still go a bit higher with reasonable voltage. Past 3.9 does start sucking voltage on the Phenom II though. But this is characteristic of almost all AMD chips since Clawhammer.

 

To be fair, the cost of i7 builds is still coming down, an i7 920 D0 and a reasonable X58 board will set you back £375 here now, when a Q9550 and high end P45 board is still £290 or so, even with the recent price drop. For the potential performance, especially if you overclock, i7 really isn't that bad. The achilles' heel is the question over how much longer it will last as a platform.
You have to remember, you're comparing AMD's second effort at quad cores against Intel's first. The Q6600, granted now has two steppings with G0, but was the very first quad core CPU Intel made, and yet as you say, is still a match for the X4 940, AMD's second best yet. What will define how things proceed from here is if Intel stagnates or not...

 

As far as reenecoding goes the tests done for benchmarks are lame (no pun intended since they only reencode a few minutes, and that really doesn't say much about how they will fair over a period that is greater that is close to 4 hours for even the fastest processors. All H.264 variants are multi pass encoders, and the most common of those used is X.264 developed by the original developers of Xvid, because it's free, fast, and well thought out. Reencode speeds tend to vary depending on the shift of bitrate in an individual movie, which means that encoders will speed up and slow down a bit during the process. I'm in contact with a lot of video heads who are using fast AMD rigs and my Q9550 is coming in on average better than an hour faster of the course of the reencode.


Here is an example of how that might work using the new AMD 6 core compared to other AMD 4 core processors.

http://techreport.com/articles.x/17005/11

When I can get my hands on a 6 core processor that scales well using X.264 (which is highly multi threaded)then I'll move to the one that is affordable.


Note that even on the second pass the Xeon 5550 takes out the 6 core Opteron 2435. So the message is that even though X.264 can run 6 threads at once the 6 core AMD still falls behind the xeon.

 

Sam,
It's only important as to it's historical perspective. It's simply the fact that it was a technological achievement. A working 45nm 4 core Microprocessor had never been successfully built and operated before, and thus it becomes a part of computer history that can't be glossed over. Even if AMD had gone broke and gone out of business, they should still get credit for what they did accomplish!

The power consumption, I wouldn't exactly call dramatic. i7, I would call dramatic, although they do idle very efficiently, but the minute you put a load on it, it's power consumption goes up to the moon. I've noticed in the last year or so that every time AMD brings out a newer version of their chips, the power consumption goes down a bit. They are far more efficient today, than they were a year ago!

Best Regards,
Russ

 

its wierd how i7 uses alot more power when i have seen D0 steppings at 4GHz of the 920 at under 1.2Vs lol

 

sophocles,
Intel didn't demonstrate anything! By their own words:

I think there is quite a bit of difference between holding something in your hand and actually running one! Nobody even saw it run! I don't think that announcing anything is the same as demonstrating a working CPU!

Russ