Remember when computers always got faster and cheaper

[VirtualLarry]

10/22/2014 Be aware going in that this is an old thread (2009) revived by the OP.
admin allisolm

This post was inspired by this Hot Deal.

It seems that due to forced market segmentation by Intel, computers are going in two directions. It used to be that computers, from all market segments, got cheaper AND faster at the same time.

Now, computers are getting cheaper and SLOWER (look at the Atom system mentioned above, slower than a Pentium 4 of yesterdays bygone era, while not much cheaper).

They are also getting faster and MORE EXPENSIVE. Witness Intel cutting off lower-end Core i7 CPUs, forcing you (in september, when Core i5 launches) to purchase an "Extreme Edition" CPU, if you want to move to the Core i7 platform (for multi-way CrossFire/SLI support).


I guess what I'm trying to say is, say goodbye to the days of relatively unified platforms, in which both the high-end and low-end shared the same architecture, in which the joe-six-pack people effectively subsidised the R&D necessary for the enthusiast-class machines.

 

[JaYp146]

Gamers drive the higher-tech sector, everyday-types drive the lower end (ie. netbooks and basic desktop PCs).

 

[Cogman]

'tis the effect of a monopoly. Why should Intel develop a newer, faster chip when nobody can compete with their current offerings? Heck, AMD has just BARELY released a cpu that is somewhat competitive the the Core2 (the Phenom2).

And you can bet that when AMD gets close to releasing a cpu that is comparable to the i7 that intel will just release a new 20% faster CPU. Until then, they will sit on the design and rack in the profits (no reason to undermine their current product offerings).

This is why you keep R&D running when you are ahead (Yes, I'm looking at you AMD) the phenom was a terrible rushed disappointment that set AMD back by a huge margin. Granted, there was no way for them to predict that the Core2 would be such a performance leap, they should have at least had a decent offering that wasn't screaming "rushed product!". Phenom 2 is what phenom 1 (minus the die shrink) should have been.

If anything I am still more angry at AMD then Intel for the current situation we are in.

As for the Atom. It is a specialty CPU. You shouldn't expect a cpu that was soley designed for low-power to be faster then the previous generation. Rather, look at the cheep celerons and pentiums (which are faster then the former generation) if you want to make a good comparison.

 

[OILFIELDTRASH]

I totally blame Intel for this. I will never spend a $1000 on a cpu. I will seriously consider an AMD system next build. If 1000 will get me comparable systems on either platform I'm going AMD. Right now I'm running an Intel build.

 

[firewolfsm]

Atom is barely faster than processors from FOUR generations ago. It's ridiculous and will hold back software development a lot.

And tech forums should all help start a movement to encourage people to buy AMD over Intel. It could make a noticeable impact.

 

[poohbear]

I beg to differ, the x3 720 is a very fast & brawny cpu and its really cheap. competition is good.

 

[Idontcare]

Originally posted by: VirtualLarry

Topic Summary: not anymore, thanks to Intel

Intel's decisions makers have a fiducial responsibility to their shareholders, not future would-be customers.

Are you honestly trying to argue the position that a for-profit business should have decision maker's at the helm who willfully ignore their fiduciary responsibilities?

Originally posted by: VirtualLarry
I guess what I'm trying to say is, say goodbye to the days of relatively unified platforms, in which both the high-end and low-end shared the same architecture, in which the joe-six-pack people effectively subsidised the R&D necessary for the enthusiast-class machines.

So it's good news for the millions of joe-sixpacks who have, until now thanks to Intel, been disproportionately footing the bill and subsidizing the top-5% enthusiast market with cheap(er) leading edge CPU's than they would have otherwise been.

Sounds like joe sixpack owes Intel a thank-you going forward, and we all owe joe six-pack a thank-you for the past 20 yrs.

 

[soccerballtux]

This does not bother me. AMD's current quads are fast enough for anything anybody currently wants to do. I do not see a quad core being necessary for gaming any time soon. By the time it is, AMD will have octo's out.
I, for one, am glad software cannot get much more bloated. Cpus will keep increasing in power exponentially, but it is exponentially more expensive to write software that grows exponentially bloated.

 

[Majic 7]

And I would like to thank AMD for charging 1000$ when they had the fastest and best.

 

[VirtualLarry]

Originally posted by: Idontcare
Intel's decisions makers have a fiducial responsibility to their shareholders, not future would-be customers.

Are you honestly trying to argue the position that a for-profit business should have decision maker's at the helm who willfully ignore their fiduciary responsibilities?

I don't follow. It's not obviously clear that this is the most profitable solution for intel.
Quite frankly, I feel that it is a mistake. With a relatively unified platform, both low-end and high-end purchasers both feed into the R&D budget, which is shared between low-end and high-end platforms (since the architecture is unified). R&D ROI is maximized.

Now, with the low-end and high-end being split architectures, they each need a seperate R&D budget, which is only fed into by purchases of their respective platforms (low or high end).

In effect, Intel has just taken out the multiplicative effect of having a shared R&D program and budget across product lines, and instead now it is only additive.

I think that this approach is more costly in the long run as far as R&D goes, and eventually it will begin to chip away at what we expect as customers as far as Moore's law goes.


Another way to put this is, for the low-end customers (Atom for example), there isn't much profit margin (granted there is slightly more volume, but go with me for this example). Meaning, that the R&D budget is going to be capped, to something lower than if there was a shared platform R&D budget. So long-term, performance improvements at the low-end are going to suffer, and Moore's law, as far as performance that is seen by customers, will suffer.

Overall, this will not benefit consumers, nor will it benefit Intel. (Unless they achive an absolute monopoly, in which case performance improvements will likely stagnate anyways.)

 

[videogames101]

Guys, atom is mobile, it's not faster, it's smaller, and more energy efficient. Not all market sectors are about speed.

 

[VirtualLarry]

Originally posted by: videogames101
Guys, atom is mobile, it's not faster, it's smaller, and more energy efficient. Not all market sectors are about speed.

Wasn't there a benchmark, where someone compared an ULV AMD64 chip to an Atom, and it had better performance/watt than Atom did? (Possibly with the system chipset power factored in too, IIRC.)

I, for one, would much rather see a single-core (dual-thread) Nehalem chip with single-channel DDR3 and integrated PCI-E, than an Atom CPU. Maybe even remove the L3 in that case too, if it's only single-core.

After all, doesn't the Nehalem arch offer the best performance/watt of any CPU currently on the market?

 

[PCTC2]

Originally posted by: VirtualLarry

Originally posted by: videogames101
Guys, atom is mobile, it's not faster, it's smaller, and more energy efficient. Not all market sectors are about speed.

Wasn't there a benchmark, where someone compared an ULV AMD64 chip to an Atom, and it had better performance/watt than Atom did? (Possibly with the system chipset power factored in too, IIRC.)

It's including the chipset. The 945GC is what decreases Atom's performance per watt. The most efficient Atoms fit within a 2W envelope. The 945GC drew like 40W. With newer chipset designs that Intel are coming out with for Atom, eliminating bloat and power requirements, it will better fit into the marketshare it was designed for: Small, portable electronics.

As for the rest of the Intel line-up, it is what is profitable, as IDC says. It does not require two completely separate R&D lines. All Intel architectures are derivatives of each other within a family. The i5 uses a very similar core to the i7, just with older DMI instead of QPI, dual-channel, and a PCIe controller. Nothing new, just a new package. Intel is moving forward, and spreading across the market. Enthusiasts have always had to pay more, and now Intel is just enforcing that, so there are less blends between "Enthusiast" and "Mainstream".

 

[Cogman]

Originally posted by: PCTC2

Originally posted by: VirtualLarry

Originally posted by: videogames101
Guys, atom is mobile, it's not faster, it's smaller, and more energy efficient. Not all market sectors are about speed.

Wasn't there a benchmark, where someone compared an ULV AMD64 chip to an Atom, and it had better performance/watt than Atom did? (Possibly with the system chipset power factored in too, IIRC.)

It's including the chipset. The 945GC is what decreases Atom's performance per watt. The most efficient Atoms fit within a 2W envelope. The 945GC drew like 40W. With newer chipset designs that Intel are coming out with for Atom, eliminating bloat and power requirements, it will better fit into the marketshare it was designed for: Small, portable electronics.

As for the rest of the Intel line-up, it is what is profitable, as IDC says. It does not require two completely separate R&D lines. All Intel architectures are derivatives of each other within a family. The i5 uses a very similar core to the i7, just with older DMI instead of QPI, dual-channel, and a PCIe controller. Nothing new, just a new package. Intel is moving forward, and spreading across the market. Enthusiasts have always had to pay more, and now Intel is just enforcing that, so there are less blends between "Enthusiast" and "Mainstream".

I'll never understand why intel saddled this uber power efficient chip to such a crappy chipset. They had much better available at the time of the release.

I really don't see what the complaint is. "Oh nos, intel is making a profit!" The comparison of the Atom is retarded. The atom is NOT meant to be intel's low end processor. It is their low power processor, And yes, there is a big difference. Check out celeron prices and speeds if you want intel's low end processors.

 

[WaitingForNehalem]

Once again everybody views Intel as an evil company and AMD as robin hood. :roll:

 

[ShawnD1]

Originally posted by: Idontcare

Originally posted by: VirtualLarry

Topic Summary: not anymore, thanks to Intel

Intel's decisions makers have a fiducial responsibility to their shareholders, not future would-be customers.

Are you honestly trying to argue the position that a for-profit business should have decision maker's at the helm who willfully ignore their fiduciary responsibilities?

Not only that but selling chips at cost would probably drive AMD into complete bankruptcy. Then Intel really would have a monopoly and we really would be screwed.

That would also mean Nvidia has a total monopoly. Double screwed!

 

[Idontcare]

Originally posted by: VirtualLarry

Originally posted by: Idontcare
Intel's decisions makers have a fiducial responsibility to their shareholders, not future would-be customers.

Are you honestly trying to argue the position that a for-profit business should have decision maker's at the helm who willfully ignore their fiduciary responsibilities?

I don't follow. It's not obviously clear that this is the most profitable solution for intel.
Quite frankly, I feel that it is a mistake. With a relatively unified platform, both low-end and high-end purchasers both feed into the R&D budget, which is shared between low-end and high-end platforms (since the architecture is unified). R&D ROI is maximized.

Now, with the low-end and high-end being split architectures, they each need a seperate R&D budget, which is only fed into by purchases of their respective platforms (low or high end).

In effect, Intel has just taken out the multiplicative effect of having a shared R&D program and budget across product lines, and instead now it is only additive.

I think that this approach is more costly in the long run as far as R&D goes, and eventually it will begin to chip away at what we expect as customers as far as Moore's law goes.


Another way to put this is, for the low-end customers (Atom for example), there isn't much profit margin (granted there is slightly more volume, but go with me for this example). Meaning, that the R&D budget is going to be capped, to something lower than if there was a shared platform R&D budget. So long-term, performance improvements at the low-end are going to suffer, and Moore's law, as far as performance that is seen by customers, will suffer.

Overall, this will not benefit consumers, nor will it benefit Intel. (Unless they achive an absolute monopoly, in which case performance improvements will likely stagnate anyways.)

All R&D projects have a diminishing returns characteristic to their scale and scope...diverting 5% of the resources from the largest project won't fatally undermine the deliverables of that project (sandy bridge will still be developed, etc) and at the same time that 5% can represent such a large absolute dollar amount that it can serve as sufficient funding to create an entirely new product with significant deliverables.

I think you do yourself a disservice to presume you know more than Intel's decision makers as to what makes the most sense for their R&D allocations to meet their long-term corporate objectives. Only the decision makers know what options were on the table, what options they downselected to (and why), and what those options are planned to actually entail five and ten years down the road.

To take a static cross-section of the deliverables from those projects as you perceive them to exist today, and then make extrapolations within the limited conceptual boundaries you possess (surely you agree you have less data than Intel does on the topic) and then to proceed to assess the resultant outcomes is simply fatally flawed thinking for all the self-evident reasons one can imagine.

 

[soonerproud]

Originally posted by: WaitingForNehalem
Once again everybody views Intel as an evil company and AMD as robin hood. :roll:

Yup, all them people with those evil i7 processors can send them my way because I am morally bankrupt and would love to have that kind of power in my rig.

 

[alkalinetaupehat]

Originally posted by: Idontcare

Originally posted by: VirtualLarry

Originally posted by: Idontcare
Intel's decisions makers have a fiducial responsibility to their shareholders, not future would-be customers.

Are you honestly trying to argue the position that a for-profit business should have decision maker's at the helm who willfully ignore their fiduciary responsibilities?

I don't follow. It's not obviously clear that this is the most profitable solution for intel.
Quite frankly, I feel that it is a mistake. With a relatively unified platform, both low-end and high-end purchasers both feed into the R&D budget, which is shared between low-end and high-end platforms (since the architecture is unified). R&D ROI is maximized.

Now, with the low-end and high-end being split architectures, they each need a seperate R&D budget, which is only fed into by purchases of their respective platforms (low or high end).

In effect, Intel has just taken out the multiplicative effect of having a shared R&D program and budget across product lines, and instead now it is only additive.

I think that this approach is more costly in the long run as far as R&D goes, and eventually it will begin to chip away at what we expect as customers as far as Moore's law goes.


Another way to put this is, for the low-end customers (Atom for example), there isn't much profit margin (granted there is slightly more volume, but go with me for this example). Meaning, that the R&D budget is going to be capped, to something lower than if there was a shared platform R&D budget. So long-term, performance improvements at the low-end are going to suffer, and Moore's law, as far as performance that is seen by customers, will suffer.

Overall, this will not benefit consumers, nor will it benefit Intel. (Unless they achive an absolute monopoly, in which case performance improvements will likely stagnate anyways.)

All R&D projects have a diminishing returns characteristic to their scale and scope...diverting 5% of the resources from the largest project won't fatally undermine the deliverables of that project (sandy bridge will still be developed, etc) and at the same time that 5% can represent such a large absolute dollar amount that it can serve as sufficient funding to create an entirely new product with significant deliverables.

I think you do yourself a disservice to presume you know more than Intel's decision makers as to what makes the most sense for their R&D allocations to meet their long-term corporate objectives. Only the decision makers know what options were on the table, what options they downselected to (and why), and what those options are planned to actually entail five and ten years down the road.

To take a static cross-section of the deliverables from those projects as you perceive them to exist today, and then make extrapolations within the limited conceptual boundaries you possess (surely you agree you have less data than Intel does on the topic) and then to proceed to assess the resultant outcomes is simply fatally flawed thinking for all the self-evident reasons one can imagine.

I love the way you talk nerdy to me IDC...



*on topic*
Really guys? Do you NEED an evil overlord to hate? I can crack out the footage from 1984 and have you go for a TMH at Goldstein.

The Atom is meant for very portable devices, which BTW have crap for processing power. Processor tech from 5 years ago would be an increase over what is currently used in the likes of cellphones, PSPs, Nintendo DS', etc. Currently what is going on in the market is a somewhat early (IMO) release of the Atom probably to prep developers for a more capable mobile platform or potentially a straightforward screw-up by Intel.

The thing with LGA1366/1156 is an effort to turn what is truly a server CPU into a desktop part. While LGA1366 is a powerful platform, it is meant for applications in which cost does not weigh heavily into it (Businesses needing powerful servers) and really is mitigated by the efficiency of the platform at the tasks it will do (which will be very multithreaded, such as multiple independent database quieres). LGA1156 streamlines this platform to be more cost efficient and better targeted towards the desktop market. You will note that the MOST EXPENSIVE chip being launched is $550(ish), netting you a processor which should compare favorably with the same $550 Core i7 which is also more expensive.

There is indeed a clear movement towards more performance for less, and what is happening now is the beginning of mobile devices more powerful than ever before, in fact powerful enough to execute the most common tasks requested by end users in general. The possibility of the majority of appliances you interact with having the processing power to interlink with each other and from there things get pretty sweet. Wait a couple years before you revolt.

 

[SunSamurai]

Originally posted by: JaYp146
Gamers drive the higher-tech sector, everyday-types drive the lower end (ie. netbooks and basic desktop PCs).

lol no.

Originally posted by: firewolfsm
Atom is barely faster than processors from FOUR generations ago. It's ridiculous and will hold back software development a lot.

And tech forums should all help start a movement to encourage people to buy AMD over Intel. It could make a noticeable impact.

Do you even know what the atom is for, dumbass?

 

[DrMrLordX]

Originally posted by: Cogman


I'll never understand why intel saddled this uber power efficient chip to such a crappy chipset. They had much better available at the time of the release.

Just to hazard a guess, but they probably did this so that existing PC operating systems would install on Atom systems without a hiccup. Maybe this was not an issue with their better chipsets but I honestly wouldn't know.

All I do know is that you can drop XP on an Atom system without thinking about it twice. If Windows 7 is indeed faster and lighter than XP at launch, then you will be able to go back and put that on there, too.

Or you can just use Linux, or whatever other PC OS you want.

 

[sgrinavi]

Including the SSDs I have about $2500 in my i7 system. Back in 2005/2006 my X2/4200 system was $2000 plus I had another $1000 in an opteron rendering engine. The I7 system is an order of magnitude faster in both modeling and rendering than my $3000 worth of hardware was.

I can't complain, I get better quality, more volume and spent less money...

 

[LoneNinja]

The 3 AM2+ systems I'm using right now were each cheaper than the gaming rig I built a few years ago on socket 754. I managed to get my Athlon X2 7750, Phenom 9850, and Phenom II X4 940 each cheaper than what I paid for the Athlon 3400 years ago. 4Gb of DDR2 ram also costs less than 1Gb of DDR cost back than too. I've got a 4670, 3650, and use the 3300 on my third machine, all of which are faster and cheaper than the 6600GT I bought years ago. Hard drives are faster and much cheaper now as well.

So I've got no complaints either.

 

[arkcom]

Both my current i7 rig and q6600 rig from a year ago were about 1/2 the price of my athlon xp 1700+ in 2003 (pretty low end at the time)

 

[SickBeast]

It's easy to say that Atom is slower, but look at all the positives it has brought:

- way cheaper laptops (netbooks)
- laptops that can run for over 9 hours on a single charge
- tiny computers that can run without a fan

In a sense it's more AMD's fault that we have these expensive computers; intel is building the best processors they can, and selling them for what the market can bear. It's not their fault that AMD is a generation behind them right now (literally; the Phenom 2 competes with the C2Q, forget about Nehalem).