Rated: 80% (1 votes)
Loading ...How relevant is Overclocking?
greyhavens24
Before anybody accuses me of AMD Fan-boism (or whatever leet speak you choose to use), this blog is not merely about whose CPU is better. Everybody’s choice will always be very subjective and based upon their past experiences. I know people that won’t buy a new Ford because they had a bad experience with a Ford Escort Mk II they had 20+ years ago. Just because they had a bad experience with a manufacturer 20 years ago doesn’t mean that the current vehicles Ford produce are equally bad, it just means that for whatever reason that persons experience of Ford was coloured by a single experience.
I do not hate Intel far from it, before the accusations fly. The current crop of chips are truly great, but I had a bad experience back in the days of Pentium and Pentium II that has made me favour AMD ever since.
One thing that is bothering me though is the flaming Phenom is getting at the hands of the CPC editorial team. I’ve mentioned it in a previous post “AMD’s rabbit out of a hat!” that I felt that CPC are less than subjective in the way the handle reviews. Don’t get me wrong, I love overclocking and try and push the limits on my CPU’s and Graphics when they are getting to the end of their lifespan (more on this later). It’s nice to get more for less.
Whilst I appreciate overclocking is what the mag is about, sometimes I feel that they neglect other aspects of a product purely based on it’s overclockability. Take for example Issue 59 (August 2008). There is a review of the Phenom X3 (For those not in the know a TRIPLE core processor) that gets a panning because it doesn’t overclock as well as an Intel.
Firstly, the chip that beats the 65nm Phenom X3 2.4 GHz 8750 (not overclocked) is the 45nm 2.66 GHz Core 2 Duo E8200. I would agree that purely on price the Core 2 part beats the AMD, but in all honesty how can you review a 2.4 Ghz against a 2.66 GHz part? It’s hardly subjective or even fair. That’s like reviewing a 1.4 Litre Escort against a 1.6 Litre Escort, it wouldn’t take a genius to realise which one would win. To further compound the lack of subjectivity they go and award a Custom PC approved to an Intel part that the review wasn’t even about, it was meant to be a comparative part in order to review the Phenom X3’s!
Instead if you compare the performance of a 65nm 2.4 GHz Phenom X3 (Triple core remember) against Intel’s closest approximation you would end up with a head to head against the 65nm 2.4 GHz Core 2 Quad Q6600. In the same issue, there is an article “Dell vs DIY” in which there is indeed a Dell Q6600 that isn’t overclocked. When you compare the Phenom X3 against the same clocked Intel Q6600 the results are quite interesting:
GIMP IMAGE EDITING:
AMD Phenom X3 8750 2.4GHz (743)
Intel Core 2 Quad Q6600 2.4GHz (826)
If my maths serves me correctly (which is a bit shaky, LOL) that represents about 8.3% quicker for the Intel Part
HANDBRAKE H.264 VIDEO ENCODING (WHICH TAKES INTO ACCOUNT MULTICORE PERFORMANCE)
AMD Phenom X3 8750 2.4GHz (1026)
Intel Core 2 Quad Q6600 2.4GHz (1263)
This represents about 2.4% quicker than a Phenom X3
MULTITASKING
AMD Phenom X3 8750 2.4GHz (660)
Intel Core 2 Quad Q6600 2.4GHz (647)
This represents about 1.3% quicker for the AMD Phenom X3 than the Intel Part
OVERALL
AMD Phenom X3 8750 2.4GHz (809)
Intel Core 2 Quad Q6600 2.4GHz (912)
This represents that the Intel Q6600 is about 5.7% quicker than the AMD Phenom X3
PRICE COMPARISON (FROM EBUYER.COM)
AMD Phenom X3 8750 2.4GHz (£123.58 Inc VAT)
Intel Core 2 Quad Q6600 2.4GHz (£131.09 Inc VAT)
Whilst you can’t argue with the fact that the Q6600 is quicker it is not that much quicker. Whether it would be noticeable in use is probably debatable. Obviously when overclocked the Intel is way in the lead. But when you compare prices, I think AMD need to bring these down in order to truly compete. All in all you have to say that AMD and Intel are pretty much of a muchness in terms of cost and performance. It will be extremely interesting to see what Nehalem has to offer as traditionally when Intel abandons a socket (or slot design if you remember them!) they kill they previous socket dead real quick! Having said that AMD have done this on a couple of occasions also, notably with Slot A and Socket 939 (more to do with the jump to DDR2 and onboard memory controllers though). Historically though AMD do tend to push a socket to the limits before abandoning it, think about how many Athlons, Athlon XP’s, Durons and Semprons you could fit into a socket A board and the fact that a good AM2 board is still compatible with the AM2+ of the Phenom.
It is also worth noting that PC Pro found that the jump from the B2 stepping (The original Phenom X4) to the B3 stepping (which by stepping designation is a very minor bug fix tweak) Phenom X4 yielded a gain of 13% in terms of performance across the range. It will be interesting to see what extra performance can be gained in future larger steppings.
I feel that CPC should be more subjective in the way that reviews are approached and place less emphasis on overclocking. When you read the sister mag PC Pro they shower praise on the Phenoms even to the point that the Phenom X4 gains a recommended award and is on PC Pro’s A-List for CPU’s (the equivalent of CPC’s Top Hardware Choices). But PC Pro’s reviews are more subjective and are based more on price to performance and they review the product not it’s overclockability.
The point to make though is that whilst Overclocking is fine and dandy and we all love to do it, it is after all a risk. As a rule I NEVER overclock unless I’m thinking about or prepared to upgrade. I have had some disastrous attempts that has resulted in a dead Opteron 146 and a very burnt out Gigabyte motherboard (we’re talking literally here!) after overclocking an Athon XP! At the end of the day, if you are overclocking there is always a risk, a risk that not everyone is willing to take.
Come on CPC if you are going to do comparative reviews lets be a bit more subjective lads!
On a different note. Is there anybody that likes the new CPC Elite list? I find it a bit pointless really, and not as good as the previous incarnation. Whilst I appreciate what CPC are trying to do, with the Budget PC, Gaming PC, Media PC and Crazy But Cool PC, what’s the point? Surely with a magazine like CPC and it’s readership demographic, we (meaning us readers) wouldn’t probably build any of those systems, instead we would “cherry pick” the components that suited our needs, i.e. Video enthusiasts would ensure that they have a fast, large hard drive, whilst gamers would perhaps skimp on the hard drive in order to free up budget for the GPU. I feel they should get rid of the systems (which take up 4 pages) and have a broader range of elite hardware individually listed. I also don’t like the fact that they’ve dropped the comments on WHY the product has been listed. Also what’s with the 2 entries for “Mid Range Graphics Card” surely there can only be one elite listed Mid range? I think they should list the hardware as Budget, Mid-Range, High-End and leave it at that!
What do you guys think?
Not yet rated
Loading ...AMD’s Rabbit out of a hat!
greyhavens24
Perhaps my previous comments in “Are we missing something?” are less than factual but here’s one post that isn’t.
Tom’s Hardware has recently reported on some hidden technology locked away in Black Edition 8000 (X3 Phenoms) and 9000 (X4 Phenoms). Apparently these processors feature “hidden pins” on the processors and chipsets that enable overclocking via the SOUTHBRIDGE which will automatically overclock the processors.
If you are running an upcoming 2.8 GHz Black Edition CPU, a motherboard with the old SB600 model (RD690) will keep the processor cores operating at 2.8 GHz. However, if you have a motherboard with the SB700 chipset, you will receive a free upgrade to 3.0 GHz. And if you get a motherboard with a SB750 chip, your processor will run at 3.2 GHz, which matches the clock speed of the Athlon X2 6400+ - the highest clocked processor AMD ever offerred.
The 780G, 790FX and 790GX chipsets all support this feature. Once AMD can unlock the potential of the Phenom this could yield a bit of an unwelcome surprise for Intel, right on the cusp of a major architecture change.
The release date of this technology is unknown, but it will be available in time for the 790GX launch.
Interesting times as I said previously. I still feel that there is an awful lot of potential to unlock in the Phenom as it is early days in the roadmap . Overclocking aside, as this is not the “be all and end all” of what makes a CPU great, when you take a look at other publications reviews of the Phenom’s they are far more favourable than CPC whose views are somewhat skewed towards the overclocking of hardware. Don’t get me wrong I overclock myself hence the reason I love CPC, but it isn’t what drives me to my purchases. CPC’s sister publication PC Pro (also by Dennis Publications) rate processors on Performance to Price i.e. Bangs for your Buck!
When you look at this the Phenom is mixing it with Intels very favourably. One of Toms Hardware’s recent reviews of Phenoms show that the the Phenom X3 8750 (2.4 GHz) is only 3.2% slower than an Intel Core 2 Duo E6550 (2.33 GHz) for a similar cost (at the time of the review), but you gain the extra core which yielded a massive 55.8% increase in speed when encoding in Mainconcept H.264 Encoder and was actually 10.8% quicker than the Intel part in Supreme Commander. Not all is as rosey though in apps that don’t utilise the additional core though with Warhammer Mark of Chaos running 21.9% slower than on the Intel E6550. Do bear in mind though that AMD 8750 is giving away 700MHz per core to the E6550 and this could represent some of the performance in the non-multicore tests. Interestinly Tom’s Hardware also notes that the AMD Phenom X3 8750 is only 14.2% slower overall than an Intel Core 2 Quad Q6600 but the Phenom is 26.4% cheaper! This is what illustrates how important “Bang for Buck” is and why the Phenom is not as bad as initial CPC reviews make it. I can’t recall one review of Phenoms in CPC that don’t mention how bad the overclocking potential is and this then makes it a poor performer.
In my experience overclockability is usually down to steppings and dates of manufacture. Historically the reason you can massively overclock is that when there is a shortage of a certain CPU resulting in faster chips being downclocked to meet a demand or similarly a manufacturing fault that means a production run is notched back a bit due to Quality Control tolerances not being neccessarily stable. This would enable us to seemingly overclock a processor to a high degree. I know this personally because I have and Athlon 64 X2 4200+ (2.2 GHz) running at 2.95GHz on air using nothing more than 268Mhz Bus and an AC Freezer 64 Pro. Another good example was some Radeon X1800 GTO’s being able to be unlocked to X1800 XT’s and gain a whole quad pipeline of vertex and pixel shaders and ROPs with nothing more than a BIOS flash. The key thing here is that SOME CPU’s or GPU’s could be overclocked to larger degrees.
Given this, I can’t help feeling that the overclockability of Intel CPU’s is a marketing ploy of Intels, a good one I’ll agree, but a ploy nonetheless. Given that virtually ALL the Intel range overclock by 25-30% I think it entirely possible that ALL the chips are in reality underclocked and that by overclocking these chips they are simply running at the speed they were intended and in reality not overclocked at all. You may well argue that this isn’t so because you have to overvolt, but the same may apply, you are actually running them at the clock speed and voltage that they were originally designed for. To a degree the Bus speed at which you overclock (and hence possible lower mulipliers) is largely governed by the quality of the hardware around the CPU i.e. Motherboard, RAM etc. In my experimenting, Lowering the multiplier had little effect on performance but a larger effect on system stability due to the fact that the FSB/Multiplier combination had less of an impact on Athlon 64 X2’s due to their Hypertransport not having a reliance on FSB frequency in the same way as Intel CPU’s quad pumped approach does.
I’m sure someone will miss the observations again and I’ll get some negative comments. Bring it on that’s what discussion is all about, but lets have some challenges to my obeservations this time not negatives about my facts. This time the facts are their for you to go and look at.
I hope this blog proves as interesting as the last LOL!
Rated: 45% (4 votes)
Loading ...Are we missing something?
greyhavens24
Firstly, let me get something off my chest. I am very much a fan of AMD’s and ATi’s, there said it! However something occurred to me over the last few months that perhaps some of the computer pundits may have missed. I’m talking about Intel’s and AMD’s swapping of CPU fortunes.
As a keen fan of Formula 1, trends have been a prevalent part of the pinnacle of motorsports for decades. In the late 90’s a Williams was the car to be in. If you were in one of those cars all you had to do was turn up and drive to ensure a podium position (barring a certain Mr Schumacher’s miracles and ahem…aggressive driving style). Where were McLaren or Ferrari? Firmly midfield. It was very much slugged out between the then Benetton Team and Williams F1. Fast forward to the late 90’s and suddenly McLaren was the car to be in with Ferrari also in the ascendency. Bring it into the 2000’s and Ferrari lead the technology race with McLaren in 2nd Place.
So what has that got to do with Intel and AMD you may ask? Well I see it as a graphic demonstration of the “law of diminishing returns”. Much of an organisations technology is based on a tried and tested winning formulation from their current technology which is evolved into the next technology that they develop. By using the technology they have, that works extremely well and tweaking the technology at each iteration, the next thing you develop makes a step forward. However whilst this works well in the short term, in the medium to long term cracks appear in the apparent dominance of the technology. Essentially the longer you continue to evolve the current technology, at each iteration, the step forward is less pronounced. Jumping back to Formula 1, this is why we see teams fall down the pecking order after a few years of dominance.
Why does this happen? Put simply, if you are leading the technology stakes there is no motivation, or need to dramatically change what clearly is working , it is just tweaked a little to improve the performance a bit further. However this plan can be easily scuppered when technological rivals develop something that the current trend setters technologies don’t have. This results in an apparent sudden leapfrog in performance over your competitors. This performance increase is usually born out of a need to rethink your current technology, reoprientate your R&D and, in extremes, abandon your once dominant technology entirely and go back to the drawing board.
So how does this relate to old rivals AMD and Intel? When you start to analyse the relative technologies of the CPU’s these arch rivals have produced there is a similar trend to Formula 1 that emerges.
It is difficult to say who produces the better CPU’s as at different times AMD and Intel have both had the technological edge and to a greater or lesser degree it is individually subjective. However it is important to note that neither seem to learn from the others mistakes. Both have made the mistake of taking a technology too far and arriving at the party too late. In both companies histories, they have both had to cull their previous, initially ground breaking designs, in order that they can adopt a new technology to push forward. In recent times this brought about both the Superb AMD Athlon 64 and the all conquering Core 2 family.
Beginning to sound a bit familiar to Formula 1 for me! AMD and Intel aren’t the only culprits though. ATi and nVidia are also doing exactly the same! Anybody, like me, reading the reviews in CPC will notice that the the jump from nVidia’s 8×00 to 9×00 is a bit of a damp squid in terms of a leap forward from the previous generation. Both AMD/ATi and nVidia are resorting to chucking a couple of GPU’s on a single card in order to eke out a performance advantage, a sign that the current architectures are at the end of their performance gains before the next step. This is nothing new both manufacturers have done this before ATi with the ill-fated Rage 128 Maxx and nVidia with the Geforce 7950 GX2.
I think that the Phenom is actually ahead of the game and it won’t become obvious until software, clock speeds and cache catch up with the all-conquering Core 2. It’s interesting that Intel’s next generation CPU - Nehalem will adopt native quad core (just like AMD Phenom) and will have an on die memory controller (just like AMD since the Athlon). By dropping the quad pumped FSB approach of the current crop of Intel processors and adopting the more direct connection like Hypertransport, Intel are stepping into the unknown and right into territory that AMD have had numerous years of experience.
The stakes are certainly high, but who will win is a tough call. AMD have been in this position before and the results was the all conquering Athlon family and ATi’s champ the X1950 XT. I can’t help feeling that AMD/ATi are just about to spring a huge surprise on a complacent Intel and nVidia, and AMD/ATi have just the heritage and pedigree to do just that.
All the signs are there, nVidia getting cosy with AMD to fight off Intel, Intel through it’s ownership of Havok Physics API siding with AMD against nVidia and Ageia GPU physics via CUDA. All in all you have to say it appears to me that nVidia and Intel are so busy watching, commenting and reacting to what each other are doing that they seem to have neglected what AMD are up to. I also think that the native Quad Core of Phenom will be unleashed when Havok starts to be integrated into the next generation of games and programmers start multithreading the code in games. I think that Ageia’s API’s days are well and truly numbered.
I also think that AMD’s aquisition of ATi was too much too soon and they took their eye off the ball and as a result Intel took the lead. But going back to my earlier comments, Intel did exactly the same thing with Pentium IV and look how long it took them to catch AMD napping. It will be interesting to see if AMD can change their fortunes and how long it takes them to close the gap. I personally think the initial signs are reasonably promising even if they can’t overclock massively.
CPC’s Clive Webster stated that we could be in for a calm 2008, I predict much of the same with the big 4 tweaking and repositioning for a very interesting 2009.
I guess I’ll just have to wait and see!
P.S. Yes I have changed the post as responses seemed to centre around my lack of F1 knowledge and personal observations about processors which were not relevant to the subject matter. I was wondering if anybody else felt that perhaps AMD/ATi are up to something as they have been ominously quiet for a while now. What do you guys think?