YABulldozerT: AMD FX Processor Prices Lower Than Expected

Page 3 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

tincart

Senior member
Apr 15, 2010
630
1
0
Why are people arguing that the Oct. 12 rumor somehow grants heightened validity to these price rumors?

If I find some random rumor about product X and then write a rumor of my own and append it, is the rumor I wrote somehow instantly more valid?
 
Aug 11, 2008
10,451
642
126
Yup, price is a bad indicator of performance when it comes to AMD, specifically since they're usually more inclined to start price wars than Intel.

Think about it, guys. When is the last time AMD priced itself at a lower or equal performance/$ ratio to Intel when introducing a new product?

This is true, and good for the user. The problem is that they can compete only on price, since the are clearly inferior in absolute performance.

They might be pricing Bulldozer low like they did the HD5000 to regain market share, or get good press, but my opinion is that the performance does not measure up.

And as far a price wars go, the only thing that is saving AMD are the regulations that prevent monopolies and selling at a loss. With all the cash they have on hand, Intel could totally destroy AMD in a price war if they were not prevented from doing so by regulations.
 
Last edited:

lol123

Member
May 18, 2011
162
0
0
Yup, price is a bad indicator of performance when it comes to AMD, specifically since they're usually more inclined to start price wars than Intel.

Think about it, guys. When is the last time AMD priced itself at a lower or equal performance/$ ratio to Intel when introducing a new product?
That would be Sledgehammer, back when AMD held the performance crown and were almost unthreatened by Intel in that respect. Which disproves your point if anything.

However I still think there's a possibility, and I hope, that the low prices are their strategy to win back market share.
 

LOL_Wut_Axel

Diamond Member
Mar 26, 2011
4,310
8
81
The i5-2400 is $189. Probably more than the cheapest 8 core BD will end up being if its that bad. The bar could be set at i5-2300 at $179.

For some years AMD has priced their CPUs based on multi-threaded performance. This time I expect it to be no different. The Phenom II X6 1090T competed with the Core i7-860 and was overall faster in multi-threaded applications, but lost in single-threaded. AMD priced it at $10 more than the 860 at $295. The 1090T is overall the same speed as the i5-2400 in multi-threaded, so they price it $20 lower to make up for its big deficit in single-threaded.

Yup, price is a bad indicator of performance when it comes to AMD, specifically since they're usually more inclined to start price wars than Intel.

Think about it, guys. When is the last time AMD priced itself at a lower or equal performance/$ ratio to Intel when introducing a new product?

Same as above. AMD prices their CPUs based on their performance, most of the time in multi-threaded. You can see this looking at the Core 2 Quad Q9400 vs Phenom II X4 940 and Core i7-860 vs Phenom II X6 1090T. Other times they have failed; for example, the Phenom 9850/9950 vs the Core 2 Quad Q6600, where the Phenom was $20 more expensive and a bit slower in everything.

If you knew the history of AMD, you'd know they price everything in accordance with performance. They're not out to give you anything free; they're a corporation. They need high ASPs and high margins wherever they can; the last thing they want is a price war with Intel, especially now that they're back to profitability.

Bulldozer isn't very different from Thuban when it comes to manufacturing costs, either. The die size is only 10% smaller and I expect 32nm yields to be worse than their extremely mature 45nm process, even if they're still good.
 

LOL_Wut_Axel

Diamond Member
Mar 26, 2011
4,310
8
81
It isn't a factual statement so it is a purely opinionated statement. In absolute performance Bulldozer wins

Sandy Bridge 12 ALU ops/12 Mem ops vs Bulldozer 16 ALU ops/16 Mem ops
Sandy Bridge 12 x 128bit int/fp Bulldozer 8x128bit fp 8x128 int

Absolute = Paper Performance

and Bulldozer simply out performs quite significantly on absolute paper specs

It's more than probably true at this point. I think only AMD fanboys are excited now regarding BD being great as an all-around performer. I've told people at many times: there's a reason why they're pricing an Eight-Core CPU against a Quad-Core and why they're giving you such high clock speeds, and it's not a good one.

They'll charge as much as they can for it. They need high ASP.

As for the rest, not sure if serious.
 
Last edited:

NostaSeronx

Diamond Member
Sep 18, 2011
3,809
1,289
136
The problem is that they can compete only on price, since they are clearly inferior in absolute performance.

In absolute performance Bulldozer wins

Sandy Bridge 12 ALU ops/12 Mem ops vs Bulldozer 16 ALU ops/16 Mem ops
Sandy Bridge 12 x 128bit int/fp Bulldozer 8x128bit fp 8x128 int

Absolute = Paper Performance

and Bulldozer simply out performs quite significantly on absolute paper specs

It's more than probably true at this point. I think only AMD fanboys are excited now regarding BD being great as an all-around performer. I've told people at many times: there's a reason why they're pricing an Eight-Core CPU against a Quad-Core, and it's not a good one.

They'll charge as much as they can for it. They need high ASP.

The same as always AMD has a track record for selling high clock/high performance CPUs that are either on par or much cheaper than the Intel counterparts @ price
 

RussianSensation

Elite Member
Sep 5, 2003
19,458
765
126
I'm going to go out on a limb here and say that there is next to no chance that 2600k is going to be beat.

AMD's top end FX-8150 is going to be compared to a 2700k not a 2600k. 2600k is either going to be discontinued or its price dropped, depending on how good BD is. A win-win for consumers.
 

LOL_Wut_Axel

Diamond Member
Mar 26, 2011
4,310
8
81
In absolute performance Bulldozer wins

Sandy Bridge 12 ALU ops/12 Mem ops vs Bulldozer 16 ALU ops/16 Mem ops
Sandy Bridge 12 x 128bit int/fp Bulldozer 8x128bit fp 8x128 int

Absolute = Paper Performance

and Bulldozer simply out performs quite significantly on absolute paper specs



The same as always AMD has a track record for selling high clock/high performance CPUs that are either on par or much cheaper than the Intel counterparts @ price

Absolutely wrong. Even looking back at the Pentium 4 vs Athlon 64, the Pentium 4 made up for its huge deficiency in IPC with extremely high clock speeds and was priced competitively. Its big problem was energy efficiency. AMD won't be able to get a lot more frequency headroom than Sandy Bridge, as seeing from their overclocking event they should only OC 300-500MHz higher on average. That won't make up for big single-threaded performance deficiencies.

For some years AMD has priced their CPUs based on multi-threaded performance. This time I expect it to be no different. The Phenom II X6 1090T competed with the Core i7-860 and was overall faster in multi-threaded applications, but lost in single-threaded. AMD priced it at $10 more than the 860 at $295. The 1090T is overall the same speed as the i5-2400 in multi-threaded, so they price it $20 lower to make up for its big deficit in single-threaded.
AMD prices their CPUs based on their performance, most of the time in multi-threaded. You can see this looking at the Core 2 Quad Q9400 vs Phenom II X4 940 and Core i7-860 vs Phenom II X6 1090T. Other times they have failed; for example, the Phenom 9850/9950 vs the Core 2 Quad Q6600, where the Phenom was $20 more expensive and a bit slower in everything.
 
Last edited:

MrTransistorm

Senior member
May 25, 2003
311
0
0
Absolute = Paper Performance

and Bulldozer simply out performs quite significantly on absolute paper specs

I'm sure you could make insanely fast airplanes out of that paper. Maybe they'll fly farther and faster than planes made from Intel's paper. It won't make a bit of difference because the only performance that really matters is when the CPU hits the socket.
 

NostaSeronx

Diamond Member
Sep 18, 2011
3,809
1,289
136
Absolutely wrong.

:whiste:

Even looking back at the Pentium 4 vs Athlon 64, the Pentium 4 made up for its huge deficiency in IPC with extremely high clock speeds and was priced competitively.

The lack of IPC was from a bad design

Bulldozer in comparison is a great design

Its big problem was energy efficiency.

From bad design

AMD won't be able to get a lot more frequency headroom than Sandy Bridge, as seeing from their overclocking event they should only OC 300-500MHz higher on average.

1.53x "nominal" stock frequencies on "normal" cooling
1.83x "nominal" stock frequencies on "exotic" cooling + cherry picked CPUs w/o cherry picked CPUs 1.65x

Exotic cooling doesn't include Dice, LN2, LHe

These are on average or the mean clock rates of various overclocks

3.6GHz 1.53x = 5.5GHz mean average
3.6GHz 1.83x = 6.58GHz mean average

That won't make up for single-threaded performance deficiencies.
Single threaded performance usually only uses 1 IPC from both mem and alu well Bulldozer has 2 of each per core
 
Last edited:

djshortsleeve

Member
Jan 11, 2011
125
0
0
Despite this I may buy a bulldozer chip just because it will work with my AM3 motherboard with a bios update. It all depends on how they perform...

AM3 in addition to AM3+? I have an ASUS M4A88TD-M/USB3 AM3 AMD 880G, will I be able to update BIOS and use BD on this?

I was waiting to buy a Phenom II x4 (prob 955) but I will also spring for a BD if I can. I want to add a solid AMD machine to my stash in addition to my 2500k box.
 

LOL_Wut_Axel

Diamond Member
Mar 26, 2011
4,310
8
81
:whiste:



The lack of IPC was from a bad design

Bulldozer in comparison is a great design



From bad design



1.53x "nominal" stock frequencies on "normal" cooling
1.83x "nominal" stock frequencies on "exotic" cooling + cherry picked CPUs w/o cherry picked CPUs 1.65x

Exotic cooling doesn't include Dice, LN2, LHe

These are on average or the mean clock rates of various overclocks

3.6GHz 1.53x = 5.5GHz mean average
3.6GHz 1.83x = 6.58GHz mean average


Single threaded performance usually only uses 1 IPC from both mem and alu well Bulldozer has 2 of each per core

Which is why AMD needs twice the cores to be competitive with Intel, correct? At least Intel could make up for their lower IPC deficit with the Pentium 4 by having much higher frequency headroom. Bulldozer does not enjoy that benefit, and putting more weak cores on a CPU is not the answer for most desktop workloads.

The Pentium 4's problem was never not being competitive in performance, but rather in efficiency.
 
Last edited:

NostaSeronx

Diamond Member
Sep 18, 2011
3,809
1,289
136
At least Intel could make up for their lower IPC deficit with the Pentium 4 by having much higher frequency headroom.

The main fault of Pentium 4 was that it was to early...and had a very poor design

The Pentium 4's problem was never not being competitive in performance, but rather in efficiency.

Price efficiency or Power(Temperature x Power consumed) efficiency or both?

Which is why AMD needs twice the cores to be competitive with Intel, correct?

Incorrect, The cores are just to help keep up with a ever growing multithreaded market(HPC, Multimedia, Gaming)

Bulldozer does not enjoy that benefit, and putting more weak cores on a CPU is not the answer for most desktop workloads.

The IPC deficit in a Bulldozer Core you are talking about is 33%(100 - 100(4/6)) lower than Sandy Bridge Core

The IPC deficit in Phenom I and II Core is 50%(100 - 100(3/6)) lower than Sandy Bridge

They aren't weak cores and it isn't the cores most people are benchmarking

SuperPi is a Floating Point Coprocessor centric benchmark and it is x87(Which happens to be very very depreciated in Bulldozer)
wPrime I'm not sure if it is core based or Floating Point Coprocessor centric if it is core based having 1 alu/1 agu less than Phenom II/Sandy Bridge can cause that, if it is the Floating Point I have no idea why it is performing bad
Cinebench is another Floating Point Coprocessor centric benchmark and it is SSE2 based, again no idea why the benchmark shows Bulldozer being slower

...and yet AMD needs 8 cores and higher clock speeds to compete with a quad-core 2500K

Please explain.

Higher Clock speeds might be do from the lack of ALUs and AGUs
(Bulldozer if compared to SB it is missing 1 Add ALU and 1 Store AGU per core)

Other than that it should perform better than the i7 2600K

Until the CPUs release and more benchmarks are used to calculate CPU performance, I recommend not using SuperPi, wPrime and Cinebench, and OBR Benchmarks to actually show CPU/Core performance
 
Last edited:

996GT2

Diamond Member
Jun 23, 2005
5,212
0
76
Higher Clock speeds might be do from the lack of ALUs and AGUs
(Bulldozer if compared to SB it is missing 1 Add ALU and 1 Store AGU per core)

Other than that it should perform better than the i7 2600K

Can I quote you on that when Bulldozer gets destroyed by a 2600K upon release?

:biggrin:
 

NostaSeronx

Diamond Member
Sep 18, 2011
3,809
1,289
136
Can I quote you on that when Bulldozer gets destroyed by a 2600K upon release?

Can I quote you on this if the i7 2700K/i7 3820 gets annihilated by the FX-8150/FX-8120 upon release?

:p

Remember I only use benchmarks that are open-sourced and compiled by an open compiler or by a microsoft compiler ;)

But, I'm not buying Zambezi, I'm waiting for Vishera(If that is the name)
 
Last edited:

exar333

Diamond Member
Feb 7, 2004
8,518
8
91
Can I quote you on this if the i7 2700K/i7 3820 gets annihilated by the FX-8150/FX-8120 upon release?

:p

Remember I only use benchmarks that are open-sourced and compiled by an open compiler or by a microsoft compiler ;)

But, I'm not buying Zambezi, I'm waiting for Vishera(If that is the name)

You will be 'chasing the dragon'. AMD's motto these days is 'wait for the NEXT CPU, it will really be good!' It's not, and then repeat...:(
 

Idontcare

Elite Member
Oct 10, 1999
21,110
59
91
Well, the future IS fusion, after all, not sure why any of us are supposed to be wasting our time buying these faux-future chips in the first place.
 

NostaSeronx

Diamond Member
Sep 18, 2011
3,809
1,289
136
You will be 'chasing the dragon'. AMD's motto these days is 'wait for the NEXT CPU, it will really be good!' It's not, and then repeat...:(

I don't know 8 cores isn't enough, I simply need more cores!

Well, the future IS fusion, after all, not sure why any of us are supposed to be wasting our time buying these faux-future chips in the first place.

Mainstream is fusion but Enthusiast will always be non fusion based
 
Last edited:

Idontcare

Elite Member
Oct 10, 1999
21,110
59
91
Mainstream is fusion but Enthusiast will always be non fusion based

If fusion fails to deliver on its promises of OpenCL enabled apps which will be GPGPU assisted then I agree because that will mean fusion turned out to be nothing more than pedestrian on-die IGP.
 

NostaSeronx

Diamond Member
Sep 18, 2011
3,809
1,289
136
If fusion fails to deliver on its promises of OpenCL enabled apps which will be GPGPU assisted then I agree because that will mean fusion turned out to be nothing more than pedestrian on-die IGP.

I don't think Fusion will fail even if OpenCL enabled apps don't appear

AMD Fusion APU is 2x cheaper(A8-3850) than Intel's Fusion(i7 2600)
(Lol, Llano is finally out of stock took till the end of September)

AMD's APU is perfect for the mainstream

Just need to get past that darn cpu microarchitecture it is like 12 years old now you need a new one AMD Gosh~
 
Last edited:

sangyup81

Golden Member
Feb 22, 2005
1,082
1
81
The ultimate goal of fusion and the products AMD have right now are nowhere near each other

Eventually the FPU is supposed to be replaced by the GPGPU.... if that can't happen, I can't really say if Fusion has a chance of succeeding long term