Forget Huron River, 22nm Ivy Bridge on the Way

Twice as efficient per clock in a lot of situations. The pentium M was the savior of intel.

UT2003 - Citadel - Min detail

640x480 (so taking the CPU and main limit)

2.6ghz dothan = 589.2 FPS
2ghz dothan = 450 FPS
3.7ghz EE P4 = 407.6 FPS

Now there was a LOT of optimisation for the P4 at the time (a lot of it suspect TBH) which hurt the PM initially, however as time went on and the practice ended we really saw how far it was ahead.

Also remember conroe is a PM derivative. PM was the genesis of it all.

 

Wow, the P6 arch must've been far ahead of it's time if a mobile CPU could out-do a desktop one. What about the Pentium-D? Was it a PM derivative or a NetBurst part?

 

Pentium-D? The one with two cores simply glued together and called a CPU?

Yeah, definitely not a PM derivative.


See:
http://www.cpubenchmark.net/cpu_lookup.php?cpu=Intel+Pentium+D+2.66GHz

 

The Pentium 4 was an example of what happens when you let marketing dictate processor design.

Intel believed that the market wanted clockspeed; while this was somewhat true, it was because Intel themselves had started the megahertz/gigahertz race. So they went for a processor that could operate at high clock-speeds. To do so, they had to lengthen the instruction pipelines. If you had a pipeline stall, you had to go back to the beginning. This led to having to enhance cache (cache misses were a real issue with the first Pentium 4 CPUs with only 256k), and work heavily on branch prediction. It also meant you had to achieve high clock-speeds to get high performance.

Scaling up clockspeed also meant heat and power demands, on a relatively high-micron process compared to today (130nm, later 90nm). Due to the heat and power demands of the P4, it made a poor notebook processor. The P4 also never reached the top speeds Intel had claimed it would in the long run.

The Pentium M took care of this by being going back to the far more efficient P6 architecture of the Pentium Pro/Pentium III family, going back to shorter pipelines and lower latency, and improving on its power architecture significantly. It was the first x86 processor to really show that executing something efficiently had a significant impact on battery life. Since the Pentium M, Intel has played the efficiency game.

As for the Pentium D, it is just two Pentium 4 processors glued together, whereas The Core Duo is basically two improved Pentium M cores, with improvements to cache architecture and some other areas. The Pentium D never came close to having the power envelope necessary for mobile computing; the Core Duo was the first solid dual-core mobile processor.

 

Awesome post, thanks!

 

Wasn't that return to more sensible decisions quickened by the fact that they were getting thoroughly trounced for the budget range to the high end? When you see a Duron that is supposed to compete with the Celeron also crush some of the middle-end P4, you know that something is wrong with your architecture...

 

On the desktop end, yes. but not with the Duron. AMD and Intel played cat-and-mouse back and forth with the Pentium 4 and Athlon XP for some time; the Athlon XP stayed competitive and was often cheaper, giving it a great foothold in the enthusiast market. The Duron crushed the Celeron in most cases because it wasn't as limited, but Intel had the upper hand in multimedia and gaming apps with the P4 when compared to the Athlon XP, and their performance lead slowly gained (though it cost more).

Then came the Athlon 64. AMD's decision to go with an on-die memory controller (among other things) allowed the Athlon 64 to keep up with (or at times, greatly surpass) Intel on a clock-per-clock basis; a 2GHz Athlon 64 3200+ could compete head-to-head with a 3GHz Pentium 4 in many areas. Intel's response was the Pentium 4 (Prescott), a heavily redesigned P4 that was supposed to be able to scale into the 4GHz range. It never made it there, and at lower clock-speeds, its longer pipelines and higher-latency cache (despite being larger) actually made it slower than previous Pentium 4 Northwood processors in some scenarios, and no faster except in SSE3 aware applications (Northwood having only SSE2 instructions).

Intel's stumble with Prescott gave AMD some time. Shortly after Intel inroduced the Pentium D (two P4 Prescott cores on one chip), AMD released the Athlon 64 X2. The X2 was a faster processor, and more energy efficient. Up until Conroe (Core 2) in the desktop market, AMD might have had the upper hand for some time, but for Intel's marketing might, capacity for production, and stable chipset platforms for their processors, which kept AMD (not always fairly by some accounts) from achieving more market-share than they might have.

As for the mobile market, the Athlon XP-M didn't gain a large foothold; Intel had the market-share with the Pentium 4, and the Pentium M started a reign of dominance that has largely held until now.

P.S. One note of interest is that due to the way software has evolved, Hyper-threaded Pentium 4 processors are still somewhat useful, though obsolete, on a low-end desktop. The Athlon XP, being single-threaded, has aged more poorly.

 

To give you an idea of how bad it was, at around 4ghz the P4 was using more power in leakage than from transistor switching.

 

Keeping that CPU in idle was more energy consuming than keeping it at 100% load. LOL

 

No, it meant idle consumption was very high.

You add leakage + switching losses to get the total power consumption.

 

Intel reveals skinny Ivy Bridge 'Ultrabooks

Haswell is when we get the goodies.

 

Just as I thought they need architecture change to drastically reduce power. Haswell(or its 14nm shrik) would be my next purchase

 

That's just marketing, they said the same about Clarksfield and there was no revolution. If they say they will reduce power consumption by half without a die-shrink then that means reduced performance as well, I am sure of that.

Keep it chill, new architecture usually boost performance by 10 - 15% at best. Die-shrinks usually have a bigger impact so Ivy Bridge is a safer bet.

 

so where's the forget ivy bridge, haswell on the way - thread?

i will forget sandybridge if i don't have a failing system, at least.

 

It dropped a few pages back in the forum.

http://forum.notebookreview.com/har...1289-forget-intel-ivy-bridge-haswell-way.html

 

man anand is on a strak regarding ivy

thunderbolt integration

a more agressive power gating leading to a tdp adaptability

 

Still not excited. Waiting for Haswell. Want to see a mainstream cpu at 15w tdp. My penryn(T9300) thinkpad is good enough until then.

 

Intel had mainstream CPUs and still has them at 25W TDP but they are not very popular.

 

P series CPUs were everywhere in the premium mid range, they were quite a bit more expensive though.

 

That is partially true. The P7350, P7450, P7550 were all well within the mid-range. Indeed they were a bit more expensive but not by far. I remember a friend of mine had to opt between a laptop with a T6600 @ 2.20 GHz and one with a P7450 @ 2.13 GHz (the rest of the specs were the same) and the price was almost the same. In the end he opted for the P-one because it offered him better battery life.

This can also been seen by looking at official Intel prices: The T9600 @ 2.80 GHz was $316 while the P9700 @ 2.80 GHz was $348. The difference in price was very small.

 

Not only better battery life, but the P-series were good because they had much lower temperatures.

Mr. Mysterious

 

Intel to launch Ivy Bridge in March 2012

 

Are we still waiting another year, or is there any more news regarding ivy bridge's release? I haven't found any.

 

EDIT

Never mind the answer is the post above you.

 

That news is also just over a month old. There may have been developements since then, that's why I asked.

 

Nothing's come out in the past month from Intel, there's been a lot of Bulldozer leakes.

 

I thought Ivy Bridge was going to be this year? Ugh. Well I mean good, it'll give AMD time to release BullDozer/Trinity.

 

Ivy Bridge will be out in March 2012 and probably the first laptops that have it will be delivered in April 2012.

 

OT, but Trinity is due out next year also, Bulldozer is due this quarter.

 

Ultrabook, new CPU structure expected to be main focuses of IDF in September

 

The whole ultrabook concept is perfect for me, AMD is also reported to have a/many Trinity chips in the 18W range. But I hope that reducing TDP won't mean that newer chips will not be much faster than old ones.

 

I hope the rumor of 12W ULV ivy bridge chips(and 7W Haswell ULV) are true. That should hopefully result 8hrs+ battery for ultrabooks plus laptops are hopefully cooler.

 

So far the only results I've seen for Ivy are 1.8GHz desktop chip with 95W TDP! no turbo. Still a little bit early IMO to make too much from it though.

IMO I would expect much the same TDP as previous chips as the CPU has to work with the hardware (mainboard) but hopefully a gain in speed or more cores. Of course there will still be the specialized chips with lower TDP for embeded applications but as far as TDP I think it will be much the same as before.

 

IDF 2011 all about Ultrabooks (and Ivy Bridge)

 

Intel pulls back from crossing Ivy Bridge
Is PC softness dragging down Intel's 22nm plans?
Intel investment reduction plans may affect PC and semiconductor players

 

See, I don't know if the delaying of Ivy Bridge makes sense to me. I feel like, if anything, it's the Ultrabook form factor that's going to invigorate sales, and while Sandy Bridge can do the Ultrabook form factor, from everything we've heard, it's Ivy Bridge that's going to do it well, with less compromises.

 

yes, 2012/2013 would be the year we are going to have large selection of notebook without HDD(SSD only), standard full day usage(~8 hours) but not the MacAir crazy price.

 

better hope ssds come down in price a lot more as intel chips need to do also. intel i think is starting to realize that arm chips are a big threat to their business and they need to consider that.

 

Why is this no surprise to me? More heel dragging from Intel, and more time until adoption of USB 3.0. I'm now officially giving up on Intel all together.

 

Personally, I don't see SSD price a big issue anymore. Have to remember that network speed is increasing as well(both wired as well as mobile) so a 100-150GB SSD(already at reasonable price in absolute $ term) is already more than enough for many situation. Not that many people need to carry several blueray with them all the time.

 

100-150GB is still expensive for most people, plus it's still not massive for videos or games.

 

It is difficult to quantify 'expensive', 'most' and why I said I can only see it from a personal perspective.

I saw a deal of Dell Vostro V131 @ 500 CDN. If I sold the HDD comes with it and replace with a M4 on sale, it would be around 700. Still a very reasonable price.

As for not massive for video or games, it again depends on individuals. Around me, I haven't seen one hard core PC gamers, they are either casual gamers like those who place angry birds or facebook games or those who plays xbox/ps3.

Same goes for videos, I see many friends have video needs but most are short clips of vacations or things like that(rather than full length 1080p Harry Potter) and the trend is to put them on the cloud and may be a NAS at home.

 

My problem is that I do need to carry around several (75 to be exact) GB around of my music. I'm aching for an affordable 256GB SSD. And by affordable, I mean, justifiably so enough that I can upgrade from my 160GB Intel G2 SSD, which I paid $400 for.

 

Asustek to launch Ivy Bridge-based ultrabooks in April 2012

 

Ivy Bridge GPU Performance: Up to 60% Faster than SNB & Better QuickSync

 

Ivy Bridge Configurable TDP Detailed

Will make processor choices even more confusing!!

 

Im seeing a rift between consumer level and workstation level processors appearing.

The ULV is going mainstream and the ''normal'' tdp ones are going to be for enthusiasts or workstation

 

Of course Intel had to play catch up on GPU performance considering that's where AMD currently leads. I've said it before, Intel only innovates when it absolutely has to, and never an inch more. I am glad that Intel is finally taking GPU performance seriously (they have to in face of AMD), but I'll never forgive them for holding back innovation and their heal dragging on the USB 3.0 issue.

 

Intel has been trying to 'tak GPU seriously' for a while. It is more about the NIH syndrome (AMD bought ATI, not doing it internally). As for they are finally within stricking distance, it has more to do with their migty fab process(so you can put everything inside one chip) and the 'overtime good enough rules' phenomina. The HD3000(or whatever in Ivy) still lacks by a wide margin comparing to top of the line ATI/Nvidia but that is only needed for lesser and lesser games. Most games would be quite playable even with HD3000.

 

I agree with you in some sense, however intel never intends to compete on the high end, not to mention AMD as well. Those gpu imbued cpus are only going as far as mainstream goes, and this should be enough for sometime.

Unless AMD dictates that it can achieve the performance of a 300W TDP chip, in a 120W CPU chip and be capable to fit all that in small package. Although this certainly automatically rules out notebooks, and this is where the market is headed.

Sincerely I found that I can game enough on the HD3000 if the performance is 60% improved on ivy, I will be more than happy, since it will take me to the same performance level as entry mid range chips of next year, or so I guess.