mufdvr3669
Weaksauce
- Joined
- Aug 23, 2008
- Messages
- 68
They are not stopping the manufacturing of desktop chips. Please read the article before commenting on what it says.
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
AMD quits
- Thread starter fightingfi
- Start date
They are not stopping the manufacturing of desktop chips. Please read the article before commenting on what it says.
DeathPrincess
Fully [H]
- Joined
- May 15, 2010
- Messages
- 18,205
Did you read the thread? Please read the thread before you come across all condescending.
DaedalusHelios
2[H]4U
- Joined
- May 19, 2008
- Messages
- 2,264
Dear AMD,
Please sell yourself to IBM so you will have the money to grow. Your current trend will end with you looking like VIA.
Please sell yourself to IBM so you will have the money to grow. Your current trend will end with you looking like VIA.
who owns most of the market in 2nd and 3rd world counties?
DaedalusHelios
2[H]4U
- Joined
- May 19, 2008
- Messages
- 2,264
If you are referring to AMD, it doesn't look like it's helping.
Didn't you copy pasta that post from Tom's Hardware ?
Unfortunately, AMD can't sell itself to anyone.
Unfortunately, AMD can't sell itself to anyone.
I'm sure there could be some legal trick to allow the sell.
intel?
mufdvr3669
Weaksauce
- Joined
- Aug 23, 2008
- Messages
- 68
Did you read the thread? Please read the thread before you come across all condescending.
Yep I have read the thread. More importantly I have read the article, and nowhere does it say they are going to stop making desktop chips.
DeathPrincess
Fully [H]
- Joined
- May 15, 2010
- Messages
- 18,205
It wasn't related to the article, it was related to other comments as you were told not long ago. E.g.
You are actually are already seeing it. You can't just look at the markets where Intel has competition to prove or disprove this point. You need to also look at the markets where Intel doesn't have competition....like it's server market.
Intel's pricing here has taken off like a airline jet. DP processors based on Conroe pretty much stayed in lock step with AMD's prices more or less all the way to the top end. With Nehalem they increased slightly. Now? Try a Quad Core @ 2.66 for $800 which is 2.5 times the price from a year ago.
Don't fool yourself. If AMD went the way of the do-do you will be paying more for less. Actually you already are. Do you honestly believe that 3.5 Ghz is the top end of what Intel's capable of? Intel could release an Ivy-based SKU tomorrow @ 4.2 Ghz. Why hasn't it? Because it doesn't need to. It doesn't have any competition to release faster models so it can release a processor at 3.5 Ghz today and wait 6 months from now to give you another speed increase tomorrow.
In addition, the grey market problem was solved years ago. That DP processor I spoke of earlier that's for 1366. If you want a new server chip that's quad core you can buy the 2011 socket one for 300.....at 2.4Ghz. Which one are you going to buy? Not to mention Intel EOL's SKUs at a very quick clip. Don't think so? Feel free to find a 1156 processor coming from a reputable source. You could always go grey market here. But you'll do without the warranty, and you might get the product you ordered.... you might not.
Then again you could spend $100 on the processor for the EOL'd socket from E-Bay, or you could pay $65 for the current socket processor and maybe another $65 for a new MB. This is how Intel keeps the grey market in check. You EOL the sockets (whether they really bring more performance or not). What do you think costs more, a new pin out or an entirely new processor architecture? Besides that Dell isn't going to go to E-bay for it's processors and no reputable OEM is going to give you a warranty on used items that there aren't any replacement items for. Intel EOL's sockets BECAUSE it has competition, not because it doesn't.
MacLeod
[H]F Junkie
- Joined
- Jul 28, 2009
- Messages
- 8,279
You can just look at their current line of processors. Want to overclock an AMD chip? Go right ahead as even their $70 chips will OC. Most have unlocked multipliers and the ones that dont you can crank on thru FSB. Want to overclock an Intel chip? Thatll be $230 minimum.
I dont think Intel will price all their processors at $1000 if AMD dies but we will certainly start seeing prices going up and performance going down. There will be no incentive for Intel to push itself and cut costs to offer more at a competitive price cause there will be no competition.
I dont think Intel will price all their processors at $1000 if AMD dies but we will certainly start seeing prices going up and performance going down. There will be no incentive for Intel to push itself and cut costs to offer more at a competitive price cause there will be no competition.
When we say go down we are talking about what intel would do vs. competition vs. without.
What performance segment?
Do you think Intel has been making performance segment processors the last several years?The reality is that AMD has to move to the lower-TD range of the market because that's where the money is. They're significantly behind Intel and treading water the past few years has only gotten them tired and on the edge of drowning. Chasing a diminishing market segment, performance segment, wouldn't bring in any money. It would take years to catch up and require billions and billions of dollars. They don't have neither the time nor the money for it (or the fabs)
The issue Read has is whether AMD have the chips for these products. Meyer and Ruiz were all about moar coars and higher performance, at least on the server/laptop lines, with little SoC style integration. AMD is selling Trinity processors for ultrathin designs yet relies heavily on fast dual-channel DDR RAM... w0t?
If Kabini/Temash offer great performance at a low cost and sip power frugally, they should do alright. If Read is left with shitty architectures to try to sell to OEMs, they're fucked. It takes 4-5 years to see a CPU design from start to finish, so if the road paved by those before him leads to bankruptcy, there's nothing he can do about it.
LAiN_OfTheWired
Gawd
- Joined
- Nov 6, 2008
- Messages
- 825
You're clearly delusional if you think Intel hasn't been making CPUs in the performance segment. They've maintained steady jumps in single-threaded, multi-threaded, IGP, and memory controller performance to continually outpace AMD (while AMD goes and shoots themselves in the foot as far as single-threaded performance is concerned to boot). They make a 3.4GhZ CPU that looks better in every usage scenario than their closest competitor's CPUs even when the competitors CPUs are overclocked upwards toward 4.6GhZ. If that isn't 'the performance segment' then I'm not sure what is.
Intel has been making 35W/45W TDP processors since ~2006. The fact that they scale up well doesn't mean they're making performance segment processors.
Notice that 40%-of-the-die GPU and single digit IPC increases (Haswell should bring double digit, but it'll likely stay the same clocks so the 5-15% performance bump will equal Ivy). Then there's the lower TDP as well.
Neither AMD nor Intel has created a true performance segment processor in years. They make server processors and laptop processors. AM3+ is server, 2011 is server. FM2 is laptop, 1155 is laptop.
http://www.anandtech.com/show/6355/intels-haswell-architecture/2
Haswell is a 17W TDP-focused architecture. Ultrabooks, essentially.
LAiN_OfTheWired
Gawd
- Joined
- Nov 6, 2008
- Messages
- 825
So you're telling me based off of laptop chips (which aren't meant to be performance chips) that Intel doesn't make things for the performance segment? You've never owned a Sandy Bridge or Ivy Bridge K-series, have you? Have you heard of them? I mean, you use the internet, right? LGA 1155 is a desktop socket. Please use Google or Bing before saying something obviously false.
Just because the majority of the market shifts one way does not mean every company no longer caters to another segment at least a little bit. I can guarantee you this 2600K is a 'performance segment' chip, and so is the 3770K. The 8-core Bulldozer CPUs were meant to be as well, hence bringing back the FX moniker. They didn't really hit SB performance, but they've got some real multi-threaded oomph to them and were targeted at the high-performance desktop segment. Again, man, Google is your friend.
Just because the majority of the market shifts one way does not mean every company no longer caters to another segment at least a little bit. I can guarantee you this 2600K is a 'performance segment' chip, and so is the 3770K. The 8-core Bulldozer CPUs were meant to be as well, hence bringing back the FX moniker. They didn't really hit SB performance, but they've got some real multi-threaded oomph to them and were targeted at the high-performance desktop segment. Again, man, Google is your friend.
They make processors for the server and mobile segments. They bin them for the desktop and unlock the multiplier for the K-versions. It does not a performance segment processor make
And for the record, you might be the biggest idiot on these forums with that bolded statement above. Apparently, a man who reads microarchitectural papers, and knows how Intel and AMD design products, hasn't ever heard of Sandy Bridge or Ivy or even an unlocked multiplier. That's your argument. Well done.
I recall you using the word "delusional" above. To whom was that actually addressed?
About that reading...
http://en.wikipedia.org/wiki/Product_binning
http://www.realworldtech.com/sandy-bridge/
http://www.realworldtech.com/bulldozer/
In fact, Intel's Conroe was based on the old Pentium-M, a mobile architecture. Their Pentium 4 was supposed to be the "performance segment" processor, which utterly fell flat on its face. So, oddly enough, it was the mobile processor that was a great performance segment processor while the actual performance segment processor sucked donkeyballs. Since then, everything Intel has made has been a mobile processor if you're looking at the desktop. It isn't until they deviated with the 1366 platform which offered the server chips instead. Bear in mind that these chips are still architecturally the same as their server/laptop brethren, the only differences are with respect to binning, chipsets, and the "features," which essentially amount to flipping a "no" to a "yes," and that includes the multiplier, ECC RAM, virtualization, etc.
-- one last edit:
I'm awaiting your response eagerly. I can only hope it's as insightful as the last one
Last edited:
What performance segment?
The reality is that AMD has to move to the lower-TD range of the market because that's where the money is. They're significantly behind Intel and treading water the past few years has only gotten them tired and on the edge of drowning. Chasing a diminishing market segment, performance segment, wouldn't bring in any money. It would take years to catch up and require billions and billions of dollars. They don't have neither the time nor the money for it (or the fabs)
The issue Read has is whether AMD have the chips for these products. Meyer and Ruiz were all about moar coars and higher performance, at least on the server/laptop lines, with little SoC style integration. AMD is selling Trinity processors for ultrathin designs yet relies heavily on fast dual-channel DDR RAM... w0t?
If Kabini/Temash offer great performance at a low cost and sip power frugally, they should do alright. If Read is left with shitty architectures to try to sell to OEMs, they're fucked. It takes 4-5 years to see a CPU design from start to finish, so if the road paved by those before him leads to bankruptcy, there's nothing he can do about it.
Yes, I do think Intel has been making performance processors for the last several years. I'm also not arguing the merits of AMD's shift in philosophy. I'm arguing what effect this change will have on Intel processors.
So Intel's mobile processors range from 75-130tdp then? Or was that just an attempt at skewing reality to mold it in a fashion that you see fit?
There's a significant difference between "think" and reality. There's no doubt that Intel has had better desktop processors, but they haven't designed a single desktop performance segment processor since the Pent 4. They design it around the 35-45W range and then bin them accordingly, whether up to desktop or down to Ultrabooks, 77/95W and 17W respectively. Haswell now approaches design with 17W in mind and will bin both up to desktop and laptop, as well as down to sub-10W levels.
Both Intel and AMD pick a certain TDP as their goal, but they allow flexibility because it has to, usually, be able to bin both up and down. For example, Trinity is a 35W chip in design but bins up to 100W (though it's likely lower, as AMD's binning is quite broad in the desktop space). On the other end, they have Bobcat, which sat at sub-20W. It can be binned and trimmed down to under 10W, as in AMD's Hondo which is a 4.5W TDP chip.
So when Intel and Anand both state that they, intel, have been designing processors for 35/45 range we should ignore it and listen to you instead? Maybe we can listen to the other guy, with his profound insights about multipliers being unlocked.
I think you're confused. Please point me to where I even hinted that you should ignore what Intel and Anand said? I'd love to see it. Let me attempt to clear up your confusion since you don't seem to know who said what, and that includes your own claims.
YOU said, everything on the desktop since conroe is a mobile processor.
Desktop processors since Conroe have had a TDP rating of up to 130watts, with a typical range of 75-130... I don't see laptops using these processors. What I'm trying to say is, you're wrong. (not Intel or Anand) and you cetainly aren't saying the same thing they are.
Then you're misunderstanding me...
The chips are designed with mobile in mind, and that's what I mean by mobile processor. They're centered around that laptop-TDP range and then binned from there, whether up to desktop or down to Ultrabooks. That makes them mobile processors in my eyes, as that is their intended market and primary focus.
Keeping mobile in mind when designing architecture doesn't automatically make the desktop variants a mobile processor. You make it seem like designing a versatile architecture means something it doesn't.
Intel doesn't call them mobile processors, Anand doesn't' call them mobile processors, they're not classified as mobile processors on Newegg.
We aren't wrong becuase we don't see things through "your eyes"
Intel doesn't call them mobile processors, Anand doesn't' call them mobile processors, they're not classified as mobile processors on Newegg.
We aren't wrong becuase we don't see things through "your eyes"
There's what I actually say and what you think I say.
Nevertheless, it's a digression. His point is they, AMD, need performance segment processors instead of Fusion/Zambezi server/laptop focused architectures. That's never going to happen on either side, whether AMD or Intel. They start designing their product within the TDP range that suits the biggest market. They're corporations who are out to make money, not please enthusiasts. When laptops outsell desktops at 2-to-1 or 3-to-1 ratios, they're not going to design anything for the performance segment. The designs have been focused around laptops for several years now.
Just because you can make a desktop out of it or because it has an unlocked processor doesn't make it a performance segment processor. You can take an Atom core and bin it to 60W, would that be a performance segment processor?
I'll just quote myself to save time...
I have an idea.... When a desktop processor outperforms everything else in it's class or "segement" i'll call it a performance processor, and you can call it a mobile processor.
One embraces reality, the other still believes Intel and AMD actually give a damn about the enthusiast. That doesn't mean they suck as enthusiast processors, but let's not pretend that's what they are. I'm sure there's not many enthusiasts bragging about how awesomely shitty their HD3000 is or how thankful they are because Intel decided to waste 30% of the die area on useless graphics.
or just plain ol' desktop segment processor, because they haven't designed a single one with that in mind since the Pent 4. Both synonyms would work, whether "enthusiast" or "performance segment" or "desktop."
1155 is desktop (enthusiast or performance segment. You can replace that as you wish) insofar that it's binned differently than laptop and has a different chipset/case. It's a "desktop" chip, originally designed for laptop, allowing enough flexibility to fit both ULV and 77W TDP.
Like I said, if an Atom was binned to 60W+, would you call that a desktop/enthusiast/performance segment processor? Because by your criteria it would be. If it fits in my desktop, it's a desktop processor? No...
Last edited:
AntiSocialMunky
Gawd
- Joined
- Nov 4, 2006
- Messages
- 589
Well they had those ridiculous $1000 conroes of unlocked multiplierness.
But there isn't an atom at 60w, so your point is well... I'm not sure really. Doesn't seem like you have much of a point. Maybe if there's ever a 60w Atom (there won't be) or when the rest of the world starts calling everything under the sun a "mobile" processor, you'll have one. Until that happens though all you have is your own opinion that everythign should be called a mobile processor, which I don't agree with, and apparently, neither does Intel or anyone of authority.
Last edited:
You are the wannabe quitter of all times. Prognosticators like you who have an obvious bias try to discredit everything they do NOT understand. And boy they do not understand a whole lot. What you have done is not strengthen any reasonable argument that AMD is about to fold. . Instead you have discredited yourself. While times are certainly uncertain for AMD, they have a lot goijng for them, it is likely if they can survive the next12-14 months that they will make a big turnaround in their business, primarily at Intel's expense. I am sure there is some venture capital money that will be coming AMD's way to tide them through the next 3 quarters. If that in case turns out to be for real, they will make it.
Acer_Sheep
[H]ard|Gawd
- Joined
- May 18, 2007
- Messages
- 1,201
You don't understand well the difference between desktop and mobile processor, technically there is no difference between those, the difference is only in different package (mobiles come in PGA package and have no heat shield) and being downgraded, binned lower, removed extra features you see on desktop ones, having less cores and HT instead and are overclocking locked, but that's all, the architecture and the silicon is same. As desktop processors are much more powerful than their mobile counterparts, they had to be designed first and downgraded to the level of mobile CPU. It's simple as that, because there is no preference for either mobile or desktop segment, they introduce new architecture that will feature both desktop and mobile processors. In the case of Atom, it is offered completely the same way both for desktop and laptop segment. Which does mean it is a desktop CPU, not enthusiast or performance however, if it would be offered in 65W version it would be at least as powerful as Celeron G530, and it that case it would be fully standalone desktop CPU competitive in the low end class.
It's been like this with desktops and laptops for so many years now, how could you were able to miss that.
That has to be one of the top ten most annoying internet cliche sayings.
Klrgoten
Gawd
- Joined
- Oct 16, 2012
- Messages
- 642
Title is in no way accurate... they are laying people off, not exiting the market... companies do this all the time
Do I win anything?
Pieter3dnow
Supreme [H]ardness
- Joined
- Jul 29, 2009
- Messages
- 6,784
This thread is something which was posted on another website. The resident drama queen at Toms hardware page thought it would be interesting yet entertaining to make up some stuff and twist his thoughts into it.
This however is not strange thing and on top of it is off topic in the cpu section. Then again whenever there is [H] review on AMD cpu you will hear it again and again like a broken record.
That's backwards. The chips are designed at 35-45W TDP then binned both up (to desktop) and down (to ULV/Ultrabooks). It's in the quote and link I already posted where it's stated by Anand and Intel at their IDC. There has to flexibility with respect to the architecture and its clock speeds, TDP and even graphics to allow for that wide a range. But make no mistake, it's planned for the 35W (dual core/low clocked quad) or 45W (quad high clocked) level.
And the above has been true since 2006, so I'm not sure what you're referring to.
Last edited: