PDA

View Full Version : Intel Gulftown/Core i9



r.e.
28-Nov-2009, 14:21
I'm trying to decide whether to buy i7 now or wait for i9. A few days ago, Tom's Hardware referenced an article written by some people who apparently got their hands on an i9 CPU, an article that has now been removed at Intel's request or demand:

http://www.tomshardware.co.uk/gulftown-core-i9-benchmarks-tests,news-32315.html
http://pclab.pl/art39718.html

Does anyone know when in 2010 i9 is supposed to be released? First quarter works for me, second quarter not so well. This is for a custom desktop build, so the timing for Dell et al. doesn't matter.

Thanks.

Ed Richards
29-Nov-2009, 07:06
The small bit I have read is that the older chip is a good buy, unless you need some special quality of the new one. I am not sure what that would be in the still photo processing world. Photoshop is not that well optimized for multiple cores, and data movement is probably a bigger deal for ultimate speed than is pure FLOPS. I am not even sure that I7 really offers much over the previous generation of chips.

r.e.
29-Nov-2009, 07:16
Thanks,

I've seen suggestions that CS5 will make more use of this technology and I also want to use this computer for video editing.

Part of what's driving my decisionp-making here is that Intel's new chips will involve new technology. I'm still trying to figure out the implications of that for motherboard compatibility, etc.

Jack Dahlgren
29-Nov-2009, 10:22
If I were you I'd wait for the i11. It will surely be faster and more capable. It would also work better with the UHD video from large format sensors.

In all seriousness, the performance gains from generation to generation are on the range of 20% or so. Not a huge deal. You are going to get more bang for your buck by going to an SSD rather than increasing CPU speed. First generation technologies take a stepping or two to shake out, so if you need something now, buy now. Whatever you buy will be surpassed shortly after you buy it anyway.

Doug Fisher
29-Nov-2009, 10:46
Here are a couple of the latest notes from Engadget:

---

Early Core i9 benchmarks promising, make you wonder why you even bothered with Core i7

By Paul Miller posted Nov 24th 2009 1:24AM
Immediate obsolescence is an age old problem in the computer industry, but it doesn't look like the upcoming Core i9 "Gulftown" processor is going to do anything to solve it. Word is from early benchmarks of the upcoming Intel processor is that it bests the current Core i7 at the top of the heap with speed gains as large as 50% -- directly in line with its addition of two cores on top of the Core i7's existing four. Of course, six 2.8GHz cores aren't quite as exponentially helpful when applied to non-optimized tasks, but with most major modern software development aimed at better utilizing multiple cores, the core overkill of Core i9 will likely prove increasingly useful over time. At the start, however, Core i9's improvements will come at a premium: 130W power consumption instead of 95W in Core i7, and of course a high-end only price tag to match. Word is we'll be seeing these chips hit the market in early 2010, possibly as soon as January.

----

Just when you thought you had enough gigahertz in your life, along come the folks at Impress to blow the doors off Intel's upcoming crop of desktop processors. In the highly detailed charts there's wild talk of a low-powered "S" version of Core i5 that lowers the chip from 95W to 82W, a new Core i3 line that strips out the Turbo Boost technology and dips into budget-priced territory, and word that at the time of this roadmap at least the Core i9 "Gulftown" chip isn't slated for until Q2 of next year. We could probably bore you all day with the details, so hit up the source link for all the sordid details before we get ourselves too worked up.

http://translate.google.com/translate?js=y&prev=_t&hl=en&ie=UTF-8&u=http%3A%2F%2Fpc.watch.impress.co.jp%2Fdocs%2Fcolumn%2Fkaigai%2F20091127_331818.html&sl=ja&tl=en

r.e.
29-Nov-2009, 11:11
Thanks for the Engadget link. I'll have to start following them.

John, I was prepared to buy a solid state drive until a marketing guy from a highly reputable company that makes them told me, over the course of dinner, to take a pass for the time being.

PenGun
29-Nov-2009, 13:42
You need an X58 chipset. The P55 stuff will not do westmere or the new gulftown 32 nm 6 core shrink.

The 1366 LGA socket is what you need for this. Don't cheap out on lynfield if you want this upgrade path. Bloomfield is what you need.

Marko
29-Nov-2009, 14:19
If I were you I'd wait for the i11. It will surely be faster and more capable. It would also work better with the UHD video from large format sensors.

Pfft, i11 will be obsolete the moment it comes out, the i13 will blow it straight out of the water. And it will be much better suited for CS7 too. :D

John Whitley
29-Nov-2009, 15:21
The i9 really just provides more cores, and some minor incremental improvements in other areas. For that, you can just build out an 8-core system now and be done with it. That nonsense about a 50% performance boost is a bit daft, since one could just as well utter "nearly a 100% performance boost for a dual i7 system over a single i7 system". Depending on Intel's pricing, dual i7 may well be *cheaper* than an i9.

Also, for your main apps (CS4/5, and video processing, it sounds like) make sure that they're CPU-bound (i.e. the CPUs are the limiting factor for performance, instead of RAM, or disk speed) and are even capable of using more than four cores. You may get more total performance simply by adding more RAM than you'd planned on, or as another poster said, by adding fast disk (SSD, dedicated scratch disks, etc.) to the system.

If your goal is really to achieve top performance with CS5, then I'd wait until CS5 is released to identify the hardware spec that will maximize performance within your budget. That may not work for you so well, since CS5 isn't due until second quarter 2010. The biggest reason is that Adobe is optimizing for both multi-core CPU use as well as for leveraging GPU resources. Determining the best balance for $$ spent requires a working crystal ball right now.

Mike1234
29-Nov-2009, 16:17
I'm seeing some darned good advice here.

paulr
29-Nov-2009, 16:51
If your goal is really to achieve top performance with CS5, then I'd wait until CS5 is released to identify the hardware spec that will maximize performance within your budget. That may not work for you so well, since CS5 isn't due until second quarter 2010. The biggest reason is that Adobe is optimizing for both multi-core CPU use as well as for leveraging GPU resources. Determining the best balance for $$ spent requires a working crystal ball right now.

Exactly. Software that can even use 4 cores efficiently is still pretty rare. Apple is trying to change this with its GCD technology, but even this requires coders to use it in their aps. It won't have much effect on current software.

Also, fwiw, apple is likely to have a short term exclusive on the 6 core chips, just as it did with the last two generations. Don't know how many months this typically is.

r.e.
29-Nov-2009, 17:19
The 1366 LGA socket is what you need for this. Don't cheap out on lynfield if you want this upgrade path. Bloomfield is what you need.

Is it clear that the 1366 socket will work with the i9?

The issue for me is Bloomfield now or i9 if it is available by spring 2010. Part of the calculation is whether the Assus PT6 motherboard, if I get it now, will work with i9.

I'm not looking for a cheap solution. I'm looking for either a clear upgrade path that involves replacing as few components as possible or, ideally, purchase of i9 if it looks like it's going to be on the market in the next few months.

Some of the people who have responded to this thread may be thinking that I'm ignoring their advice. It isn't that, it's just that I've been round the block on this stuff and what you are saying doesn't relate to the question that I asked. However, it may be very useful to others. All I want to get a handle on is compatibility of current motherboards with i9 and to see whether anyone here has a handle on likely release of i9.

On the suggestion that Apple will get access to i9 before custom PC builders, gamers, etc., if there is any truth to that, I'd like to know. It may have an impact on my decision. The suggestion sounds daft to me, but maybe I'm missing something.

PenGun
29-Nov-2009, 18:07
Is it clear that the 1366 socket will work with the i9?

The issue for me is Bloomfield now or i9 if it is available by spring 2010. Part of the calculation is whether the Assus PT6 motherboard, if I get it now, will work with i9.
There are no guarantees but Intel has said the x58 chipset will support Gulftown 6 core 35mn CPUs on the 1366 LGA. It should just be a BIOS upgrade.

PenGun
29-Nov-2009, 18:16
Exactly. Software that can even use 4 cores efficiently is still pretty rare. Apple is trying to change this with its GCD technology, but even this requires coders to use it in their aps. It won't have much effect on current software.

Also, fwiw, apple is likely to have a short term exclusive on the 6 core chips, just as it did with the last two generations. Don't know how many months this typically is.

Yeah but then you would have to put up with a Mac. They seem to charge more and more for less and less. Still if you love em' ... you love em'. ;)

John Whitley
29-Nov-2009, 19:04
On the suggestion that Apple will get access to i9 before custom PC builders, gamers, etc., if there is any truth to that, I'd like to know. It may have an impact on my decision. The suggestion sounds daft to me, but maybe I'm missing something.

There is a rumor floating around the 'net to this effect, which may have originated in this Hardmac.com article. (http://www.hardmac.com/news/2009/10/15/future-mac-pro-apple-to-enjoy-short-term-exclusive-use-of-future-xeon-cpu) It is well documented that Apple has made exclusivity arrangements with component vendors in the past, so there's precedent. More grist for ye olde decision mill...

r.e.
29-Nov-2009, 19:21
Well that site's brief suggestion is a long way from saying that "apple is likely to have a short term exclusive on the 6 core chips, just as it did with the last two generations. Don't know how many months this typically is".

What is the basis for this statement, if any? I don't see any discussion about an Intel/Apple deal to this effect, let alone one calculated in months. On the face of it, it doesn't make sense. If what paulr is saying about the history is true, I'm interested in knowing about it. It may lead me to buy an Apple, but this is the first time that I've heard this suggested.

Donald Miller
30-Nov-2009, 00:53
I rode to the airport with an Intel engineer last week. He said that they have CPUs with 15 cores in the pipeline. So, as others have said, if you want the optimal performance wait awhile. Intel not only designs and develops the CPUs and ram but they also design the MB for their processors. This MB design is then given free of charge to companies like ASUS for their implementation in MB fabrication. I have the I7 in my latest desktop and it works very well for video processing...I would think that video processing (HD) would represent a lot larger data load than any still image.

Donald Miller

Eric Leppanen
30-Nov-2009, 11:17
http://www.tomshardware.com/news/intel-roadmap-core-i5-i7,9167.html

"Roadmap also shows a dual-core Clarkdale-based Core i5 with Hyperthreading that'll launch at the faster clock speeds starting at 3.43GHz and ramping all the way to 3.73GHz."

Trying to guess what Intel will do in the future can rapidly become self-defeating. In my opinion, you should upgrade your hardware now if you feel you can achieve a significant, cost-effective increase in performance with your current software. If your goal is a system optimized to CS5, you should wait until CS5 comes out and then make hardware decisions.

The problem with LGA1366 versus LGA1156 based systems is that it is very hard to document any meaningful performance improvement with LGA1366, aside from a very few specialized applications and multi-GPU gaming apps. It is not clear if this will change any time soon, particularly if the above roadmap turns out to be legitimate.

And the SSD market is still in its infancy, with significantly improved products expected over the coming months. Personally, I find not having an SSD-based system disk annoying, but it does not heavily impact my productivity at this point. Therefore I am waiting for the technology to improve (TRIM support, lower cost, faster write performance, etc.) before plunging in.

Mike1234
30-Nov-2009, 12:06
Ehh... just wait until your system is 10 years old then upgrade to "nearly the best" available. The difference will knock your socks off. :) I say "nearly the best" because "the best" is always overpriced regarding cost vs. performance value. I'm exagerating a little on waiting 10 years, BTW.

PenGun
30-Nov-2009, 12:12
X58 with the 1366 LGA is the high end board. P55 with the 1156 LGA is for the masses and is pretty near as fast but will see less memory and has a different CPU architecture.

Even the cheap 920 i7s overclock very well and 3 G + is pretty easy to achieve.

P55 will _not_ run the 6 core gulftown.

Kirk Gittings
30-Nov-2009, 12:44
Ehh... just wait until your system is 10 years old then upgrade to "nearly the best" available. The difference will knock your socks off. :) I say "nearly the best" because "the best" is always overpriced regarding cost vs. performance value. I'm exagerating a little on waiting 10 years, BTW.

My strategy exactly, but i wait about three-four years.

Mike1234
30-Nov-2009, 13:03
I agree, Kirk. Unless one is making money at it and staying busy enough to need the added speed/efficiency to justify the expense... or simply has deep pockets... there's no reason to upgrade until the overall performance (speed and/or size handling) at least doubles. Three-four years between upgrades and staying just under the "best available" radar seems about right. Save your money to buy better lenses and more film. :D

Bruce Watson
30-Nov-2009, 14:53
I agree, Kirk. Unless one is making money at it and staying busy enough to need the added speed/efficiency to justify the expense... or simply has deep pockets... there's no reason to upgrade until the overall performance (speed and/or size handling) at least doubles. Three-four years between upgrades and staying just under the "best available" radar seems about right. Save your money to buy better lenses and more film. :D

Meh. I bought excellent glass and film before thinking about a computer. That's why this computer is from 2002 (that's right) and the monitor is failing. But at least I think my priorities are in order. Now if only I can coax another year or so out of it... :eek:

r.e.
30-Nov-2009, 15:25
My strategy exactly, but i wait about three-four years.

I just happen to be at three years :)


P55 will _not_ run the 6 core gulftown.

That's my understanding too. What's less clear to me is whether an X58 PT6 will run it.

brad martin
30-Nov-2009, 15:40
I agree, Kirk. Unless one is making money at it and staying busy enough to need the added speed/efficiency to justify the expense... or simply has deep pockets... there's no reason to upgrade until the overall performance (speed and/or size handling) at least doubles. Three-four years between upgrades and staying just under the "best available" radar seems about right. Save your money to buy better lenses and more film. :D


My thoughts exactly.