Ivy Bridge 3700K @ 77W TDP

[toyota]

How is 95W and 130W not random? You take the improvements that's given.

Like I said 77W is what 10% voltage reduction with square in relation to voltage results in. You can reach 95W with 2600K if you use power virus code.

I meant to say odd not random.

 

[Idontcare]

How is 95W and 130W not random? You take the improvements that's given.

Like I said 77W is what 10% voltage reduction with square in relation to voltage results in. You can reach 95W with 2600K if you use power virus code.

LOL, so true. The TDP numbers are arbitrary in the sense they could have been 96W or 94W or 95.5W, etc.

Why 77W and not 80 or 75? Just because

 

[Nemesis 1]

I would also point out the pre chart is showing for the 77 watt part / In the Added section . 8 threads . So what exactly does this mean ? As its aligned with the already existing 2600k that already does 8 threads. So what exactly does that mean ? Poorly made chart that should have said added 8 threads aligned with the 2500k I5 ?

 

[bunnyfubbles]

remove the GPU and it would look like we're ready for 130W TDP 8C/16T IB-E chips (or 6 core 98W)

 

[ElFenix]

uhhh... 77W disappointing... ur aware how many cores are on it right?
This aint a dual core cpu... u know how hard it is to shrink TDP, while increasing performance, even though a node reduction.

77 is right in line with what i was expecting (75).

i guess my c2q can wait until april probably. i got other things to buy as well.

You're doing it wrong. SB-E is 32nm on a larger die. It isn't third generation, that's like saying Bloomfield was SB.

that's his point. SB-E is getting a 3 for the first digit, not a 2.

 

[Idontcare]

77 is right in line with what i was expecting (75).

i guess my c2q can wait until april probably. i got other things to buy as well.



that's his point. SB-E is getting a 3 for the first digit, not a 2.

I don't mind Intel taking the TDP-tiers down a notch provided it doesn't mean they are engineering these things to be weenies that can't take the heat of being OC'ed to the point that they are throwing out 250W of power consumption.

 

[Nemesis 1]

Ya I doubt that will ever happen .

http://news.softpedia.com/news/Intel-Ivy-Bridge-and-Haswell-CPU-Overclocking-Detailed-227629.shtml

 

[Nemesis 1]

Haswell

http://news.softpedia.com/news/Intel-Haswell-CPUs-Could-Pack-Multiple-GPU-Cores-227894.shtml

 

[jpiniero]

I was expecting 65 W actually (80 W -> 45W cpu with same speed as 2700K/3.5/3.9t, 15 W -> 20W gpu with double performance+)

Intel is too scared to release a processor over 4 Ghz.

 

[bunnyfubbles]

I was expecting 65 W actually (80 W -> 45W cpu with same speed as 2700K/3.5/3.9t, 15 W -> 20W gpu with double performance+)

Intel is too scared to release a processor over 4 Ghz.

lol scared of what? certainly not of AMD, which is exactly why we haven't seen a 4+GHz CPU from them

no need to ramp up the clocks without the competition

 

[Nemesis 1]

I was expecting 65 W actually (80 W -> 45W cpu with same speed as 2700K/3.5/3.9t, 15 W -> 20W gpu with double performance+)

Intel is too scared to release a processor over 4 Ghz.

I thought intel already released a 4ghz server chip . I thought I read that here.

 

[GammaLaser]

Yes, if you count the Xeon E3-1290 which turbos to 4GHz. I think there may be another one that also turbos that high.

 

[Concillian]

my question is why such an odd number like 77?

I imagine they set the original spec the same as SB, but when they got the end result at the performance they were looking for, they were like, "hey, this is way lower than spec, we can drop the spec to look better" so they did and 77W was where it ended up.

A bad way to set a spec: let's use this nice round number, it sounds good.

The right way to set a spec: We got XW from our average part and we have a gaussian distribution showing a process sigma of Y. We'd like to have no more than 1 PPM out of spec unless something is F'ed up in our process... that means the upper end of our spec should be ... 77W. Cool, so our spec is 77W.

We have both kinds of specs where I work. The latter way of doing it results in funny numbers, but they're realistic, achievable and make sense. The former way is some dude's pet number and often results in internal politics, wasted resources, and is more likely to generate issues at the customer level because that dude just pulled the number out of his ass instead of looking at what we can actually make.

 

[tatertot]

I was expecting 65 W actually

65W with no clock increase would be:

1.05 / (65/95) = 53% perf/W improvement!

Those are some high expectations.

 

[Caza]

It appears Intel's 22nm process is coming along nicely...

Top bin Ivy desktop 4C/8T only uses 77W, down from 95W Sandy.

A bit different than GlobalFlounderings.

It seems until AMD fesses up to the design problems in Bulldozer, they will keep throwing GFlo under the bus. With Bulldozer having over 2 billion transistors (2x as many as Sandy Bridge) there's no way it can keep at the same TDP as Intel on a clock for clock basis. Those transistors need power. They would have to clock it under 2Ghz to get there.

 

[StrangerGuy]

It seems until AMD fesses up to the design problems in Bulldozer, they will keep throwing GFlo under the bus. With Bulldozer having over 2 billion transistors (2x as many as Sandy Bridge) there's no way it can keep at the same TDP as Intel on a clock for clock basis. Those transistors need power. They would have to clock it under 2Ghz to get there.

BD is drawing the same power as its stated TDP 125W at stock.

While SB actual power draw at stock is nowhere even close to its TDP.

 

[ElFenix]

I was expecting 65 W actually (80 W -> 45W cpu with same speed as 2700K/3.5/3.9t, 15 W -> 20W gpu with double performance+)

i think that's asking too much from a node shrink.

 

[Dadofamunky]

man 2012... when 4ghz overclock means nothing but squat....

:\

i remember when 3.6ghz was almost viewed as godly...

4.6 GHz OC is the new 3.6 GHz OC!

 

[bunnyfubbles]

I thought intel already released a 4ghz server chip . I thought I read that here.

Yes, if you count the Xeon E3-1290 which turbos to 4GHz. I think there may be another one that also turbos that high.

Technically the jpiniero claimed that intel was too scared to release chip that was over 4GHz, and even a chip that turbos to 4GHz (but not over) wouldn't qualify for that

4.6 GHz OC is the new 3.6 GHz OC!

aigo did say 3.6GHz as viewed as "almost godlike", which would have been back in the day of 65nm Core 2 Quads where you had to get really lucky to get a chip that could do that, especially on a quad like the Q6600. Today I'd say 4.6GHz is fairly common place, and that 5GHz would be the analogous near-godlike mark to hit

 

[psolord]

So...will we see air cooled 6Ghz clocks on Ivy or not?

 

[bunnyfubbles]

So...will we see air cooled 6Ghz clocks on Ivy or not?

we haven't even breached 5GHz out of more than 1% of current chips (and of those who do claim to have a stable 5GHz rig, most are running dangerously high volts for 24/7 use) and you're asking not just about 6GHz, but air cooled 6GHz

considering Intel capped the multiplier at 63 I highly doubt it, just because some SB chips are able to hit 5GHz doesn't mean we're suddenly going to leap forward another GHz, those really are the rare exception, just like there are an exceptional few 45nm chips that could do 4.4-4.5GHz

when we look at each process node, the realistic upper ends for stable computing on reasonable air cooling for quad cores have been:
65nm = 3.2-3.4GHz
45nm = 3.8-4.0GHz
32nm = 4.5-4.7GHz

I think we'll be lucky to see more than a 700MHz jump from SB's current practical range, so anywhere from 5.0 to 5.4GHz is where I'd predict practical air cooling will limit a majority of IvyBridge CPUs. Golden chips and/or high end cooling would likely net much faster clocks likely into the 5.5GHz range, but I think it will still be a pipe dream to expect 6GHz outside of supercooling and redonkulous volts

And if we really consider things like the fact that SandyBridge wasn't Intel's first foray into 32nm (they had a year to brush up on it with Westmere) and that Ivy will pop Intel's 22nm cherry, I think we should really be hoping IvyBridge gives us at least an easy 5GHz to at least break that milestone more than anything, it wouldn't be outside the realm of possibility for there to be a flaw somewhere along the line that severely limits Ivy's clocking potential beyond that of SandyBridge.

but yeah, i'd be surprised if we hit 6GHz on common air even by Haswell and believe it to be more likely that it will take Broadwell or Skylake to achieve 6GHz as a common milestone

of course I could be totally wrong, I'm just basing my guesses off of what we've seen from the past. Its always possible a new architecture throws us for a loop (although I doubt we see anything as drastic as Netburst from Haswell or Skylake) but that would only bring us to another point - would 6GHz even matter if its IPC isn't as fast as we're accustomed to? Right now another ~10% in IPC + a "guaranteed" 5GHz out of IB would be a solid upgrade from what SB "guarantees" us. Personally, I think 5.5GHz would be amazing, although I'm expecting something more along the lines of 5.2-5.3GHz (which would put it right in line with being 1GHz slower than its multiplier cap, much like how a realistic high end for Sandy is 4.7GHz, 1GHz slower than its 5.7GHz multi cap). But yeah, if IB can give us 6GHz on air I'd probably die from an epileptic nerdgasm.

 

[greenhawk]

reading though that, it does appear some things I will not be 100% happy with, but still looking to upgrade to IVB.

77W TDP, down from the 95W(?) TDP. Good, but that looks like 100% of the improvements from going to 22nm and 3D transisters is being used for getting lower power. No faster chips it appears.

77W vs 95W, about 19% reduction.

Going to 3xxx, a little funny seeing as the 2xxx was ment to mean second generation of this family of cpus, though having SB-E starting on 3xxx is a step away from that approach of naming.

 

[greenhawk]

but when are they going to bring out a 6 core CPU for the high end desktop, that is not priced outrageously?

Just to add, the current 6 cores are about $600 and $1000.

SB and IVB will not get a 6 core, but on the off chance it does (so taking sales from the SB-E range that intel wants you to buy into), I would expect the IVB 6 core to start at $700, no were near the $400 (which would be "reasonable" given a cost per core count just being extended out to 6 cores).

 

[greenhawk]

And its supposed to cost $294 or something, cheaper than 2600K. The 4 core clocks are better than 2700K with 3.5GHz Base/3.6GHz Turbo 4 core.

I would hope it would be a little cheaper than the 2600K seeing as the cheapest SB-E is a locked cpu. Proberly more than a standard 2600 though. Ofcourse, the only reason it is so priced is to get people into the SB-E market so their only option for future upgrades to faster CPUs is $600 or $1000.

 

[greenhawk]

I would also point out the pre chart is showing for the 77 watt part / In the Added section . 8 threads .

Most likly a poor translation. The features are being improved appon with the following is how I am reading them. Otherwise if you follow up from the lowest, the high end cpu has several threads more than 4 cores, and a massive amount of cache.