It's not just a shrink. They are also introducing high-k metal gate transistors.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
What's all the hype with Penryn?
- Thread starter pgseye
- Start date
- Sort by reaction score
It's not just a shrink. They are also introducing high-k metal gate transistors.
actually, the random write speeds are slower. A 10 second google search disproves your comment.
Yes, the random write speeds of a magnetic platter HDD are slower than a SSD...
The major delay in a traditional hard disk is the needle moving. A SSD does not have this problem, and can read any bits with equal ease and speed.
I'm still sitting on this very high and precarious throne.
EDIT: OK, half of me is on the floor. Wikipedia says I'm wrong.
But, no matter. The other half of me remains on the chair, because m point was that the MBP is not going to get an SSD (At least not that the vast majority of users would benefit from having) for a few years yet.
And it's not.
EDIT 2: (Below) Beat you! By one minute! It says there, "Last Edited by 'me' at 9:10PM"! And your post is at 11 past! HA!
And there is the point that it is Wikipedia...
EDIT 3: Damn, that edit just erased the proof that I beat you. : (
gah
must i do everything for you?
Slow random write speeds as erase blocks on SSDs generally are quite large, they're far slower than conventional disks for random writes.[2]
http://en.wikipedia.org/wiki/Solid_state_disk
must i do everything for you?
Slow random write speeds as erase blocks on SSDs generally are quite large, they're far slower than conventional disks for random writes.[2]
http://en.wikipedia.org/wiki/Solid_state_disk
As one of the guys who designed opteron/athlon 64, I take slight issue with this. Leakage might "just" be leakage, but it is an increasingly sizable portion of the overall power consumption of the part. It's quite reasonable to expect that if Intel hadn't changed their gate design that leakage would have accounted for 30% of net power.
Also, the reason the power is decreased as process size is decreased is not because "the electrons don't have as far to go." As you decrease process size, leakage goes up, wires get shorter (hence less capacitance to charge as signals switch) which reduces switching power, etc. Normally you also reduce voltage, which reduces device switching power (if you keep switching times constant).
And as you reduce voltage, all else being equal, you increase leakage, because you can't shut the gate all the way off.
And if they hadn't, the leakage currents would have been tremendous. Every time a fab does a shrink, they change process parameters. This is a fairly big change this time, but, as with all such changes, they maintain the status quo more than make a major advance.
Again, if they had just shrunk the devices, they wouldn't have achieved the full benefit of the shrink because with normal transistor structures and materials it was getting to the point that the devices never shut off, and leakage (static power) was getting ridiculous.
20% over what? 20% is about what you would expect going from 65nm to 45nm, all else being equal. No way this adds 20% over that. If anything, it allows Intel to get close to the expected shrink benefit, and that's all.
With every few generations there have been these supposed "major shifts" (that is, shifts that lay-people consider to be huge wins) that have done little other than allow Moore's Law to continue. SOI, high germanium doping, low-K dielectrics are all things that represented massive shifts in the way chips are manufactured, but none of them resulted in a performance gain beyond the normal generation-to-generation shift.
Other shifts that got a lot of press but resulted in nothing are compound semiconductors, diagonal (or "x") wires, paralyne or polyimide ILDs, etc.
Revolutionary changes are required now and then just to get evolutionary performance benefit.
You aren't going to see more than the normal 8-10% actual benefit for the first penryn chips (the ones that end up in MBP early next year). Yes, these changes will allow penryn to eventually scale to faster/less power than current chips, but that's always the case anyway.
I wouldn't be surprised if they put in a dual SSD as well as the faster chips. Right now, the 2.6s still have not shipped if they get delayed until Dec I'd be inclined to wait until January to at least see.
Frankly I would be surprised if they but a dual SSD in there.
1: 128GB SSD drive costs $6000 on newegg, and I don't think that Apple will be able to get them for significantly cheaper, and can then utilize them without making the MBP cost $8000
2: They are the same size as current SATA notebook drives... as the current design of the MBP does not allow for 2 of those drives, and apple is probably unlikely to make the next design LARGER, then where exactly would they put this second drive?
I suspect it will be an option (a single SSD). But it will be VERY expensive. Dell charges $500 for a ****** 32gb and it goes up from there. Bearing in mind its constantly getting cheaper, and it'll probably be $100 cheaper by the time I post this.
I highly doubt a SSD would be stock.
The regular macbooks and macbook pros have no chance of seeing that technology any time soon (next six months or so). The costs are too prohibitive at the moment.
However, the fabled "mini" mbp has a chance of getting this. Assuming the sub-notebook has a reasonable chance of coming out next year, I think it would be the only candidate for it. The need for weight and power reduction are crucial in a sub-notebook. In a standard notebook, there are other obstacles that play a bigger role, and as such, SSD drives wouldn't offer the same advantages relative to their cost.
I think it will be a while before standard notebooks will have SSD as a standard option.
Agreed. Just because a technology exists, and just because some laptop somewhere has it, doesn't mean apple will stick it immediately into a MB or MBP. Apple tends to be aggressive with new hardware technology only when there is no other solution or the technology can be incorporated in an affordable manner.
Indeed. Apple approaches technology in a very practical manner. The new technology must provide a clear advantage to design before Apple leaps on it.
Which is why it is going to be a while before Apple puts either BlueRay or HDDVD into their systems, simply because neither is on top yet. HDMI is not a clear cut winner yet either. Sure it is nice for hooking up to a TV, but you still need DVI for hooking up to most modern monitors. One thing I do not know is if HDMI can be converted to VGA. Cuz if not, the DVI is still the way to go. With DVI I can go to VGA, DVI, or HDMI.
That whole mess has a while to go!
Seriously though, I don't see a standard emerging for at least another year (probably more). This Christmas will determine a lot.
With Montevina, I've read that there'll be chipset support for blu-ray and HD-DVD. What does this mean exactly?
The press and blogs are more than happy to parrot Intel's marketing points, but seldom do you hear them explain Shroedinger's Equation, and how it applies to power consumption
I just remembered another "revolutionary" thing that was going to make CPU's (pick a number) percent faster/more efficient - self-timed/clockless logic.
Oh, and add Transmeta's stuff to the list, too.
Not much.
Hi.
Despite the TDP being the same, I believe the penryn processors would still run at lower temperatures. This is certainly the case for the QX9650
Review show that the penryn-core QX9650 runs at lower temperature than the QX6850 despite both having the TDP of 130W.
http://www.xbitlabs.com/articles/cpu/display/core2extreme-qx9650_13.html#sect0
QX9650 Datasheet (PDF)
QX6850 Datasheet (PDF)