Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / Uri was right...
1 2 Previous Next  
- - By Banned for Life (Gold) Date 2007-12-05 18:12
I had a discussion with Uri a while back where he backed the position that optimizing for high core count was not useful for the great majority of people, while I thought that core count would be increasing dramatically in the next year or two. It turns out, Uri was right and I stand corrected. Sales figures show that quads make up 5% of Intel's current mix, and Intel has stated that this will only go up to 7% by the end of next year. This is a much slower uptake than I had imagined. Also, Intel's octal core processor has been pushed back into 2009. A lot of this probably has to do with AMD's demise.

The takeaway here is that the great majority of people will be using single or dual core machines at the end of next year (as Uri predicted), with most enthusiasts still running quads and occasionally dual quad core machines, and rarely quad quad core machines or better. Support for more than 8 cores will be necessary only for unlimited engine tournaments, freestyle, and correspondence chess.

Parent - - By Mark (****) Date 2007-12-05 20:06
I'm also surprised at the 7% figure for the end of 2008.  The ordinary quad cores (Q6600) at the major retail stores aren't that expensive.  Do you think we'll see anything better than the Q6600 at the major stores in 2008?
Parent - - By M ANSARI (*****) Date 2007-12-05 20:49
I can assure you that anyone and everyone who is interested in computer chess is thinking of getting or already has a Quadcore system.  The price of a Q6600 powered system can be had for as low as $800 ... so it is hard to ignore the impulse to upgrade if you haven't already done so.  Add to this the fact that with a little bit of tweaking you can dramatically improve the performance of your system ... it makes the reason to upgrade even greater.  Chess gains a tremendous amount from MP support (for engines that support it) and it is one application that simply cannot have too much power ... unlike most other applications that really don't gain that much.  So while you won't find that Quadcores have a large market in general ... I would bet that a much higher percentage than 5% of the chess enthusiast crowd will own a Quadcore system by next year. 

When I travel I somtimes take some playchess games my computer has played and lost and try to improve the lines ... my Core 2 Duo running 2 cores at 2Ghz feels extremely slow ... and I feel I can get dramatically much more work on my overclocked Quad core system.  Analysis with a powerful computer is a real joy.
Parent - By Mark (****) Date 2007-12-05 21:19
What I was hoping to find out was whether the major retail stores will offer anything better than a Q6600 at 2.4 gHz in the next year.  (I'm more comfortable buying a computer from a local store, rather than mailordering it.)
Parent - By pawformation (*) Date 2007-12-05 22:01
M Ansari, Can you point me in the right direction for getting an overclocked quad core? I am not looking to build one myself. Thanks,
Parent - By RFK (Gold) Date 2007-12-06 04:15 Edited 2007-12-06 04:28
Hi M Ansari,

I've been watching the prices of the Intel Core 2 Duo E6700 stay relatively static at around $290-329-
Where as the  Intel CPU-CORE2DQ6700  $700

Intel SL9UL Intel Core 2 Extreme Edition QX6700 2.66GHz 8M $648

Intel BX80562Q6600 Core 2 Quad Q6600 Processor $280

Like you say there is a lot of room to move around in if you have the money.

But, if you are going to go that far why not throw in the ASUS P5E3 at around $350.

Then you might want some good memory sticks -Corsair XMS3 DDR3 (TWIN3X2048-1333C9) $350.

Cooling system-

Video cards


list grows ... Wow! I envy anyone who can do this and just to play chess-that is really living!!! Hat's off to you-God Bless you.
Parent - - By NATIONAL12 (Gold) Date 2007-12-05 23:13
i want to why we pay in the uk 50% more than you do in the USA this applies all over europe
Parent - By Luckybaer (**) Date 2007-12-06 12:48
Maybe your tax code?
Parent - - By Nelson Hernandez (Gold) Date 2007-12-06 18:19
This is an excellent question pertaining to ingrained socialist thinking on the Continent (in countries that weren't part of the Warsaw Pact).  "From each according to his ability, to each according to his need" is the operative Marxist ideology.  I was just reading an article yesterday about the brain drain in Denmark due to a 63% tax bracket for people making over $70,000 per year.  I suppose that doesn't even include VAT and a whole host of other absurd taxes.  For example, if we paid for petrol what you guys do, and seem happy and content to pay, there would be another American Revolution.  And what do you really get for all that taxation?  In France, a vast and restive underclass and an officious overclass of elites who manage the welfare state and protect the status quo.  In Scandinavia, no one being allowed to prosper, everyone leveled to a common denominator.

Aux armes, citoyens!  Formez vos bataillons!
Parent - - By Sesse (****) Date 2007-12-06 18:37
I live in Scandinavia, and I prosper. :-)

/* Steinar */
Parent - - By Banned for Life (Gold) Date 2007-12-06 19:40
With only 3-1/2 million Norwegians, lots of North Sea Oil, very high oil prices, and a program to invest a significant portion of the proceeds, the common denominator in Norway is pretty high...

Parent - - By Nelson Hernandez (Gold) Date 2007-12-06 20:08
Lucky Norwegians!  Now all they need to do is secede from the EU, disband the welfare state, drop taxation to near-nothing, keep out low-life foreigners like you and me, and start breeding like rabbits.
Parent - - By BB (****) Date 2007-12-07 11:15
Lucky Norwegians!  Now all they need to do is secede from the EU,

Umm, Norway is not an EU member state... (rejected both in 1972 and 1994)
Parent - By Nelson Hernandez (Gold) Date 2007-12-07 17:42
WISE Norwegians! 
Parent - - By Sesse (****) Date 2007-12-07 12:44
FWIW, we're about 4.7 million (your estimate was not that far off, though).

/* Steinar */
Parent - - By Banned for Life (Gold) Date 2007-12-08 13:47
Actually, that's a pretty big error.

Looking at the "all knowing" CIA factbook, I see:

  4,627,926 (July 2007 est.)

The most impressive thing here is that it purports to give the population in July down to the last Norwegian. This is a frequent problem with the CIA. Always very precise, but rarely very accurate! :-)

Parent - - By Roland Rösler (****) Date 2007-12-08 14:24
Oh I see, you aren´t very happy with your CIA. Maybe because of some announcements of them in recent times? :-)
Parent - By davidwhite (***) Date 2007-12-10 11:17
Roland, very observant of you to immediately pick up on the gratuitous slap at the CIA.
Currently, it's in the Administration's interests to undermine the credibility of all of our Intelligence gathering agencies,foremost the CIA;though not directly since that might provoke the security declassification of documents whose classification
has thus far provided the White House with,at least, a basis for denying their release to Congress.

Now,all Administrations,regardless of Party,have been less than forthcoming in this area but some have more to hide than others and the current one,I believe, has more than any of its predecessors to deny the light of day.

This is where the stalwarts among the general population play a pivotal part.With no direct link to the Administration, they're free tp purvey the attacks technically off-limits for the Administration.
In doing so they flatter themselves that they're involved in the 'moving and shaking' scene when,in fact,they're but puffed-up tools and often have no idea of how they're actually being played.

Incidentally,the unreliability of CIA Intelligence situation estimates has for more than 50 years been well known at the Executive level and relying on them to justify our invasion of Iraq or the level of our bellicosity toward Iran has
brought the practice of disengenuousness to a never before imagined level.

Parent - - By Banned for Life (Gold) Date 2007-12-10 12:56
Actually, the CIA is only one of 16 agencies that contribute to the NIE, which is produced by the Office of the Director of National Intelligence. The CIA's role in the NIE report was significantly downgraded in 2005 as a result of the findings of the 911 commission. The organization of the ODNI is shown below:

Remember, the intelligence community was completely surprised by nuclear weapons tests by India and then Pakistan in May, 1998, and they know much less about Iran's program. As you probably know, Iran, a country with immense oil and natural gas resources, currently has to ration gasoline because of a lack of refining capability. Rather than expend resources to alleviate this problem, they are going all out to build a nuclear power plant (that will generate plutonium) and other plants for uranium enrichment. I don't think you need to be a genius to see where this is all going.

Parent - By turbojuice1122 (Gold) Date 2007-12-10 13:26
Yes, I think that you'd be hard-pressed to find anyone, whether in the U.S., Iran, or anywhere else, who truly believes that the Iranian nuclear program is for peaceful purposes.
Parent - - By Roland Rösler (****) Date 2007-12-10 15:34
The problem with intelligence community for the public is always, what can you believe. For example, I don´t believe that CIA was surprised by nuclear weapons tests by Pakistan, a dear friend of USA for many years (unlike India).
In Iran, you have to pay 7 (EU-) cents per liter gasoline at the filling station (I had tears in my eyes, when I heard that). The gasoline price is state-subsidized. And rationing (only! 100 liters per week) was an attempt to limit the subvention. I heard it in German tv and read it in newspapers. Maybe you are right, but I´m not sure.
Parent - By Banned for Life (Gold) Date 2007-12-10 16:55

As you noted, the CIA wasn't completely surprised by Pakistan's test (they noted the final preparations (encasing the device in a concrete tube prior to the test), but they seemed surprised about the timing (i.e. that Pakistan was able to do their test only a few weeks after the Indian tests). The Pakistani tests resulted in the US imposing significant sanctions which included non-delivery of some P-3 aircraft that I was very familiar with.

The important point here though is that you don't necessarily want to give a false detection (i.e. assuming the presence of WMD when there is none present as in Iraq) the same weight as a missed detection (e.g. India's nuclear weapons test).  My personal belief about Iraq is that they had no active WMD programs in 2003 because the US bombed the crap out of them in 1998 under Bill Clinton (the smart way to deal with this kind of problem, as opposed to the moronic approach of sending in ground forces). The Iranian program would be both easier and more difficult to neutralize. Easier, because the Iranians haven't attempted to shield their facilities by placing them near schools, hospitals, and mosques, and more difficult because the Iranians have built their facilities underground and have them protected with Russian anti-aircraft missiles.

Here is the news article from the BBC in June, after Iran started general rationing of gasoline:

As indicated, Iran imports 40% of its refined gasoline due to a shortage of refining capacity. They also burn off most of their natural gas because they lack the infrastructure to capture and transport it (both wasting huge amounts of energy and making a significant contribution to CO2 emissions)! Both of these issues could probably be addressed with the capital being spent to produce fissile materials (for generating electricity :-)).

As you point out, the significant fuel subsidy is a problem for the country, which is in a serious recession even with current energy prices. As you can see from the article though, this move doesn't make sense from a political standpoint since it seems that it pissed off just about everybody (the article states 100 liters per month).

Anyway, I think the picture here is pretty clear. Iran wants its nuclear program so badly it is willing to forgo improvements to the infrastructure of its most important industry and alienate its own population to move it forward. We can only speculate as to whether this stems from a desire for nuclear weapons, or the desire to produce the world's most expensive nuclear power.

Parent - By NATIONAL12 (Gold) Date 2007-12-06 22:08
we have large numbers of people moving to spain where costs are reasonable.the trouble is its brains going out and rubbish comeing in. i see the signs of a recession due to left wing goverments planning.
Parent - - By Gaмßito (****) Date 2007-12-05 20:57
I am convinced that a great majority doesn't need a Quad machine and neither can pay those high prices.  
It is not so difficult to understand, so Uri prediction is normal for me.

Parent - - By Harvey Williamson (*****) Date 2007-12-05 21:05 Edited 2007-12-05 21:10
I agree. If you want to dominate on the CB(Playchess server) then get Rybka with the best hardware you can afford and make your own book. If you wan something to play against and analyse your games with then any Computer + any of the top engines will do. Any of them will beat you and your opponents - pick the one whose style you like the best! And maybe more important one you can understand the moves and suggestions from.
Parent - - By George Tsavdaris (****) Date 2007-12-05 22:37
Exactly. If someone is not a GM or not playing correspondence and just preparing some analysis, any normal=slow computer with any good program will do.
I recently used Pepito 1.59 in a slow laptop of 2003 of a friend and played some games against some Chess players in my village. These people had a complete ignorance about Chess programs and they only knew the mighty Fritz. When Pepito beat them all easily they were astonished by it asking about it details, if it's better than Fritz etc.... And Pepito is just a prehistoric engine running in a prehistoric CPU.

But people will always seek for the best so the best engine and the best hardware will always be the goal.... :)
Parent - By NATIONAL12 (Gold) Date 2007-12-05 22:49
to the average person the best prog is the one which gets most publicity,the general public assume that if mainsteam newspapers report a computer match it must be the srongest available.chessbase is very very clever.
Parent - By Carl Bicknell (*****) Date 2007-12-05 23:07
I think for a quick "check over" after a leauge game where the engine gets only 5-10 seconds per move I would be prepared to invest in dual core + tablebases but beyond that is pointless.

I really think that having an engine with a style of play you wish to copy is essential for improvement. I found this and am hoping for a grade of 183 ECF this season, up about 20 points from a few years ago.
Parent - By Uly (Gold) Date 2007-12-06 00:29

> Any of them will beat you and your opponents - pick the one whose style you like the best!

You bring up a very important point. Since with perfect play, chess is likely to be a draw, then in most positions several moves draw (So, all of them are 'best' moves), what the engine chooses is a matter of style. So if you like a move and your engine tells you that it's 0.25-0.50 points worse than the 'best' one, you may want to consider checking it with other engines as well, maybe you'll find that your move isn't bad at all. That's the main reason I prefer to analyze with Rybka Winfinder than with Rybka.

Don't let an engine change your playing style! As Harvey pointed out, even the moves coming from one of the weakest engines would beat your opponents (Unless you're a +Master) so you're better off using the engine that plays most like you (only, stronger) for analysis than by using the strongest engine, if playing her moves lead you to positions that you don't like.
Parent - - By Mark (****) Date 2007-12-05 21:12
The more I think about it, I guess I really shouldn't be surprised at the percentages.  Many of the people I know that are very knowledgeable about computers and use them all of the time, don't even know what processor is in their computer (and there is really no need for them to know).
Parent - - By Luckybaer (**) Date 2007-12-06 12:55
I mean, outside of computer chess, file compression and some other niche applications - maybe some graphic applications? - what "average" consumer benefits from multi-core?  If I were not on the geeky side of the spectrum, I'd probably be happy with some mid-range PC that was dual core.  The benefit would be multi-tasking, but that would be it.

Until gaming software shifts heavily into applications that are multi-threaded the percentage of customers that jump on the quad- or above bandwagon will continue to be very low.  I'll qualify that:  The percentage will be low, unless a strong push is made to market comparably priced quads (Q6600) to the masses.  I guess that would take someone with market share like Dell.
Parent - By Banned for Life (Gold) Date 2007-12-06 15:17
I think you're exactly right. There are way too few applications supporting MP to justify more than two cores for the average consumer. Two cores is justified mainly by the heavy resource utilization of Microsoft Vista which should have its own core to run on to prevent it from slowing other applications to a crawl.

Parent - - By RFK (Gold) Date 2007-12-06 01:26
Hi Alan,

Excuse my ignorance, when you say Intel's Octal core processor, you're not referring to the 45nm are you?

Parent - - By Banned for Life (Gold) Date 2007-12-06 06:02
Intel has an octal core processor on its rode map, but it won't be available till 2009. It will be a 45 nm part designed for use in servers and will almost certainly be very expensive.

Parent - By RFK (Gold) Date 2007-12-06 16:50

I apologize for asking that question-it was lazy of me. I googled it got the info.Doh! But, thank you for taking the time to answer it.

Parent - - By Vasik Rajlich (Silver) Date 2007-12-06 11:02
It's really amazing how fast this AMD vs Intel thing changed. In 2005 AMD was king on the top end, having released the first dual core, etc. Anyway, if nothing else, in 2003-2005 AMD at the very least did the customers all a huge favor.

Parent - - By Banned for Life (Gold) Date 2007-12-06 15:52
AMD vs Intel in 2004 - 2006 makes an interesting case study:

AMD was actually already stone dead by late 2005 although most analysts didn't pick up on this until late in 2006. Being significantly behind Intel in process technology while having to pay royalties to IBM and manufacturing fees to Charter, along with scale advantages that Intel enjoys from having a much larger market share, meant that AMD had to maintain a significant architectural advantages over Intel just to stay even. While AMD was coasting with the K8 architecture, Intel killed off NetBurst and based on the successful Pentium M, put the Israeli design team in the top dog position. At the same time, AMD failed to raise significant much needed capital while they were the darlings of Wall Street in 2005 and early 2006 and instead overpaid for ATI with borrowed money.

At this point, its not clear how AMD will survive. This isn't the first time they've been in the position of having to sell chips that are more expensive to make and lower performance than Intel's for less money, but in the past they weren't nearly as deeply in debt as they are now.

The good part of this is that if you shorted AMD stock two years ago, you made a lot of money. The bad part is that AMD's weakness has already allowed Intel to take its foot off the gas and this will no doubt be accompanied by higher processor prices. Without competition, Intel needs to innovate only fast enough to give people a good reason to upgrade their computers every 3-5 years and they can focus more on their gross margins.

Parent - - By Nelson Hernandez (Gold) Date 2007-12-06 18:22
Between your chip manufacturer business analysis and your dissection of the electricity woes in California several years back I am amazed at how informed you are.  You must be a regular reader of Business Week!  Bravo.
Parent - - By Banned for Life (Gold) Date 2007-12-06 20:11
As you know, making large bets for and against companies is a strong incentive to research them thoroughly. If your investment performance is unsatisfactory, it doesn't really help to blame all of your problems on Dick or Hillary.

Parent - By Nelson Hernandez (Gold) Date 2007-12-07 17:41
But it will be so much fun to blame Hillary for everything!
Parent - - By Nick (*****) Date 2007-12-06 18:37

> AMD vs Intel in 2004 - 2006 makes an interesting case study:

Totally agree, it would make a fine book to read (if you like that sort of thing).   I was an AMD "fan boy" for some time, it was only when Jobs dumped PPC in favour of Intel that I knew the writing was on the wall for them.

They need to be rescued by some kind Angel (no I am not starting a rumour).  Still, they will be in the fastest supercomputer ever (4x the biggest Blue Gene) next year when the IBM Roadrunner (Skynet?) goes online :)
Parent - - By Banned for Life (Gold) Date 2007-12-06 19:59
I was a big fan of AMD's too until it was clear that they had no follow on to K8 and no response to Intel's Core 2.

AMD received a $622M investment a little while back from a government backed group in Abu Dhabi and they've already lost about $50M of it. I still worry about China accepting a multi-billion dollar loss in exchange for acquiring AMD's CPU technology, but this is becoming less of a threat every day.

It takes several years to design and build a supercomputer, and even today, AMD's chips have advantages when it comes to interconnectivity. Unfortunately for AMD, this is a niche market (but a profitable one).

Parent - By Luckybaer (**) Date 2007-12-07 01:43
A friend who works at Intel told me, "You will never go back to AMD."

He says their product roadmap is amazing.

I am saddened, because I think Intel was pushed to their current heights by AMD's challenge.  If AMD dies, Intel will probably not pump out the great technology as quickly - nor at such competitive prices.  :-(
Parent - - By Vasik Rajlich (Silver) Date 2007-12-08 17:00
My googling skills are failing me right now, but I've read somewhere that each industry tends to have room for two. You should aim to be #1, or to be #2, or get out (which can mean moving sideways to a related industry).

It's not clear though that this always applies to high-tech stuff.

I'm not following the AMD thing really closely but it sure would be a pity if they had to concede the entire CPU market to Intel. It's not like they're hopeless - they're just behind right now and it looks like they'll stay there for the next product cycle.

Parent - - By Banned for Life (Gold) Date 2007-12-08 22:04
One of Jack Welches golden rules when he was running GE, was that if a product line wasn't #1 or #2 in its category, it should be divested. He had very strong arguments for this belief and it caught on in a lot of other places.

AMD's fundamental problem is that there are significant advantages of scale in the business they're in, and they're getting larger all the time. Being substantially smaller than Intel, heavily in debt, with higher cost production and an inferior product doesn't bode well for their future. They will need a minor miracle to survive to the next product cycle. This will become more clear in the near future when they give Hector the boot.

Parent - - By Roland Rösler (****) Date 2007-12-09 02:33
I´m rather sure, that you are right about the hard future of AMD. But this doesn´t mean, that the Intel guys can sleep very well. If AMD goes down and for example Samsung takes the rest, this isn´t a fine scenario for them.
To be #1 or #2 in the market was a fine idea from Jack Welch. But you always have to consider, if you are in the right market. On the long run, even Intel has to think about it. If there is no real opponent in their market, they will become weaker!
Parent - - By Banned for Life (Gold) Date 2007-12-09 02:46
Intel probably doesn't want to be all alone in the PC CPU market. Not that they wouldn't like to be able to relax a little and charge higher prices, but there is always the pesky EU to deal with. Before you know it, they would be getting huge fines for being successful, just like Microsoft.

Better to have a weak competitor with 5-10 % of the business. Unfortunately, even if Intel eases up, it will be tough for AMD to make enough money to cover the costs of new fabs and paying back all the money that they've borrowed.

Parent - By Roland Rösler (****) Date 2007-12-09 03:33
... but there is always the pesky EU to deal with
Ignore them! Oh, I would like it so much, if Microsoft makes an example of EU. We don´t accept adjudgement and forfeit. No further products to EU states. We hold our contracts, but no new one with EU and their population.
Okay, Microsoft has calculated (and the EU also) and we have heard nothing like this. But now, they have this parasite in their neck! It´s a shame!
Parent - - By Vasik Rajlich (Silver) Date 2007-12-10 17:17
Aha, yes, thanks.

Every business has significant advantages of scale. Are these really so much bigger in CPU production that there isn't room for a #2?

Parent - - By Banned for Life (Gold) Date 2007-12-10 18:36
Intel has some significant competitive advantages that make life very difficult for AMD and have also kept other potential competitors from jumping into this rather profitable business area:

1) Intel recently completed 45nm Fab 32 in Chandler, Az. It cost $3B. This is doable for Intel (market cap $163B) but its a real challenge for AMD (market cap $5B). AMD can't really afford state of the art fabs and must rely on outside fabs (i.s. charter) to supplement their production (probably at a cost of ~$10 per die). Being fabless doesn't look like a good fit for the CPU business area, so probably the minimum viable size for this type of business would be a market cap of ~$30B.
2) Intel has a large advantage in the process technology necessary to produce dies at 45nm that run fast at low power without requiring exotic processes. AMD (and everyone else) is way behind in this area. AMD is paying IBM to license their SOS technology, but this has not worked as well as people had hoped (and adds both processing expense and the expense of paying a license fee to IBM). IBM's inability to get this right was a major reason for Apple going to Intel (Motorola's PPC division, then under current AMD CEO Hector Ruiz, didn't even try to compete). The process technology advantage isn't a pure advantage of scale, but having 20 times as many engineers and being able to experiment at a lot more fabs certainly doesn't hurt.
3) Intel's much larger design team allows them to work on multiple projects at once (primarily multiple CPUs, multiple chipsets (some with embedded graphics), and communication chips. AMD must rely on partners for all but the CPUs for right now with embedded graphics coming soon.
4) Intel's much larger marketing effort helps it in a number of areas. One example would be the very successful Centrino brand, which showed that many customers liked the ability to get the whole solution from one vendor. AMD's attempts to copy this success have been weak.

All of these point to the requirements of the need for significant size and the willingness to continuously make huge investments in production facilities in order to be competitive. There's certainly enough money in CPUs for a significant #2, but barring unforeseen events, it seems unlikely that this will be AMD for much longer.

Up Topic Rybka Support & Discussion / Rybka Discussion / Uri was right...
1 2 Previous Next  

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill