New around here? Register your SA Forums Account here!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Locked thread
Agreed
Dec 30, 2003

The price of meat has just gone up, and your old lady has just gone down

I wonder if anyone at Intel feels silly for having done that, or if it'll just be at worst meaningless to their bottom line and at best promote enthusiasts to switch from 2600K to 2700K for no reason, or bring in more people in the in-between now that the in-between exists. Hyperthreading really does have performance benefits for everything thanks to how well modern operating systems integrate it and the architecture's efficiency at implementing it, fewer wasted cycles is a pretty cool trick. If the 2600K gets close enough to the 2500K it could start being a decent recommendation for more general usage after all instead of "get this if you do time-sensitive content creation or a lot of rendering and not for any other reason."

So AMD's switched gears already, before dropping this turd of a processor on the market, to hyping up its improved successor? Generously assuming that they can improve the performance dramatically enough to make it competitive, how many years will that have been between the announcement of the new architecture and an implementation on the market that isn't non-competitive junk?

Adbot
ADBOT LOVES YOU

HalloKitty
Sep 30, 2005

Adjust the bass and let the Alpine blast
If the 2700k is just a 2600k with a slight clock bump, you'd be an idiot to buy a 2700k.

2600k becomes the new 2500k recommendation, as you said. Where does AMD fit into this? It doesn't :(

Bob Morales
Aug 18, 2006


Just wear the fucking mask, Bob

I don't care how many people I probably infected with COVID-19 while refusing to wear a mask, my comfort is far more important than the health and safety of everyone around me!

Cao Ni Ma posted:

The 2700k will replace the 2600k so the 2600k will get a natural price cut. I'd guess they'll lower the price of the rest of the chips to compensate as well but there's a chance Intel wont do it cause lack of competition etc.

Why would they? If the 2500k still beats Bulldozer in 90% of stuff, why sell them at N - $50 when they can keep selling them at N ? Plus they can keep selling the 2600k at N + $100 and the 2700k at N + $200 or whatever.

Current prices (pulling out of my rear end) + sales (also being pulled from my rear end):
code:
Chip - Price - Sales # - Sales $
2500k - $200 - 50,000 - $10,000,000
2600k - $300 - 20,000 - $6,000,000 
2700k - $400 - 10,000 - $4,000,000
$20,000,000 in sales
Let's say the knock prices down, you end up with:
code:
Chip - Price - Sales # - Sales $
2500k - $150 - 50,000 - $7,500,000
2600k - $250 - 20,000 - $5,000,000 
2700k - $350 - 10,000 - $3,500,000
$16,000,000 in sales
They'd have to sell 25% more chips to make up the $4,000,000 difference

Agreed
Dec 30, 2003

The price of meat has just gone up, and your old lady has just gone down

Thought they already announced (back when there was still some doubt as to Bulldozer sucking) the 2700K taking over the 2600K's price point. If it changes nothing else, it's supposed to bump that down. They could say "did we say that? Oops, typo, meant ON TOP OF the 2600K, hah, weird us right?" because there's no reason not to, I guess.

Factory Factory
Mar 19, 2010

This is what
Arcane Velocity was like.
They could also just stop selling the 2600K.

Agreed
Dec 30, 2003

The price of meat has just gone up, and your old lady has just gone down

Factory Factory posted:

They could also just stop selling the 2600K.

They better not :mad:

WhyteRyce
Dec 30, 2001

A possible explanation for some of the lackluster leaked performance numbers

http://www.xtremesystems.org/forums/showthread.php?275786-AMD-FX-8150-Bulldozer-finally-tested&p=4969164&viewfull=1#post4969164

quote:

Actually, we already have such an issue known for Bulldozer, and NO bench-marked system has the patch installed!

The shared L1 cache is causing cross invalidations across threads so that the prefetch data is incorrect in too many cases and data must be fetched again. The fix is a "simple" memory alignment and (possible)tagging system in the kernel of Windows/Linux.

I reviewed the code for the Linux patch and was astonished by just how little I know of the Linux kernel... lol! In any event, it could easily cost 10% in terms of single threaded performance, possibly more than double that in multi-threaded loads on the same module due to the increased contention and randomness of accesses.

Not sure if ordained reviewers have been given access to the MS patch, but I'd imagine (and hope) so! Last I saw, the Linux kernel patch was still being worked on by AMD (publicly) and Linus was showing some distaste for the method used to address the issue. One comment questioned the performance cost but had received no replies... but you don't go re-working kernel memory mapping for anything less than 5-10%... just not worth it!

quote:

This patch provides performance tuning for the "Bulldozer" CPU. With its
shared instruction cache there is a chance of generating an excessive
number of cache cross-invalidates when running specific workloads on the
cores of a compute module.

This excessive amount of cross-invalidations can be observed if cache
lines backed by shared physical memory alias in bits [14:12] of their
virtual addresses, as those bits are used for the index generation.

This patch addresses the issue by zeroing out the slice [14:12] of
the file mapping's virtual address at generation time, thus forcing
those bits the same for all mappings of a single shared library across
processes and, in doing so, avoids instruction cache aliases.

It also adds the kernel command line option
"unalias_va_addr=(32|64|off)" with which virtual address unaliasing
can be enabled for 32-bit or 64-bit x86 individually, or be completely
disabled.

This change leaves virtual region address allocation on other families
and/or vendors unaffected.

Bob Morales
Aug 18, 2006


Just wear the fucking mask, Bob

I don't care how many people I probably infected with COVID-19 while refusing to wear a mask, my comfort is far more important than the health and safety of everyone around me!

I wonder if it's similar to the memory node interleaving issue with the Opterons:

http://www.anandtech.com/show/4486/server-rendering-hpc-benchmark-session/6

Longinus00
Dec 29, 2005
Ur-Quan

WhyteRyce posted:

A possible explanation for some of the lackluster leaked performance numbers

http://www.xtremesystems.org/forums/showthread.php?275786-AMD-FX-8150-Bulldozer-finally-tested&p=4969164&viewfull=1#post4969164

Is this on LKML? Do you have links for the patch?

GRINDCORE MEGGIDO
Feb 28, 1985


How long until NDA lifts?
Sad to see the trainwreckery so far!

The Masked Unit
Feb 10, 2010

Sir, I challenge you to fisticuffs!

wipeout posted:

How long until NDA lifts?
Sad to see the trainwreckery so far!
Rumor has it that the NDA lifts tonight.

Longinus00
Dec 29, 2005
Ur-Quan
I found the thread that has the final patch versions (that I know about). Included are kernel build benchmarks.

https://lkml.org/lkml/2011/8/5/171

Killer robot
Sep 6, 2010

I was having the most wonderful dream. I think you were in it!
Pillbug
I don't want to get hopes up too much, but some sort of terrible and correctable cache problem like that would at least explain how it ended up even behind the existing generation. I wondered if something like that was possible.

Bob Morales
Aug 18, 2006


Just wear the fucking mask, Bob

I don't care how many people I probably infected with COVID-19 while refusing to wear a mask, my comfort is far more important than the health and safety of everyone around me!

Killer robot posted:

I don't want to get hopes up too much, but some sort of terrible and correctable cache problem like that would at least explain how it ended up even behind the existing generation. I wondered if something like that was possible.

There was a software patch called FASTVID for the Pentium Pro that fixed an issue where it ran DOS games slower than the plain Pentium chips.

Agreed
Dec 30, 2003

The price of meat has just gone up, and your old lady has just gone down

That your portfolio talking? :ohdear:

This is 2011, I guess maybe given the new and unique nature of the architecture it really is possible that there is some serious OS miscommunication... Intel's Hyperthreading works so well because operating systems integrate it as well. But a last minute save that turns it from poo poo to fine? When they've already, as previously mentioned, shifted gears to talking up the successor, before this even hits?

I'm still dubious but god drat it I'm not going to be a total pessimist here, I'd love for Intel to have some real competition and frankly I think the module idea is a neat one, if it works, which I guess is what they're trying to fix.

Ragingsheep
Nov 7, 2009
Passmark reports performance between a 2500k and 2600k. I haven't kept up to date with what all the "good" benchmarks are supposed to be.

Bob Morales
Aug 18, 2006


Just wear the fucking mask, Bob

I don't care how many people I probably infected with COVID-19 while refusing to wear a mask, my comfort is far more important than the health and safety of everyone around me!

It's a 6% increase so it's not exactly an equalizer but it is a sizable improvement.

Alereon
Feb 6, 2004

Dehumanize yourself and face to Trumpshed
College Slice
Just for anyone wondering, final reviews will be up at 12:01AM Eastern, which is in about an hour.

Star War Sex Parrot
Oct 2, 2003

Alereon posted:

Just for anyone wondering, final reviews will be up at 12:01AM Eastern, which is in about an hour.
I can't wait to see this disaster unfold.

incoherent
Apr 24, 2004

01010100011010000111001
00110100101101100011011
000110010101110010
Someone put up a review in spanish

http://www.ozeros.com/2011/10/review-amd-bulldozer-fx-8150/

Star War Sex Parrot
Oct 2, 2003

TechSpot: http://www.techspot.com/review/452-amd-bulldozer-fx-cpus/

HardOCP's articles show up on their homepage, but 404 when you try to view them.

Factory Factory
Mar 19, 2010

This is what
Arcane Velocity was like.
Tom's Hardware: http://www.tomshardware.com/reviews/fx-8150-zambezi-bulldozer-990fx,3043.html

Engadget has some useless "Hey, lookit these marketing slides!" drivel.

E: As I'm reading, it seems that all of the FX CPUs are the same silicon, just with Bulldozer modules turned off? Jesus, way to leverage your lovely fabs.

Factory Factory fucked around with this message at 04:18 on Oct 12, 2011

freeforumuser
Aug 11, 2007
All these official reviews simply proves beyond doubt BD is a fail of epic proportions, in performance and especially power draw (MY GOD, extra 200W from stock to 4600MHz at full load?!)

As seen from: http://www.guru3d.com/article/amd-fx-8150-processor-review/7

freeforumuser fucked around with this message at 04:21 on Oct 12, 2011

Rawrbomb
Mar 11, 2011

rawrrrrr

freeforumuser posted:

All these official reviews simply proves beyond doubt BD is an fail of epic proportions, in performance and especially power draw (MY GOD, extra 200W from stock to 4600MHz at full load).

Are these dev chips or final chips? Could it be something with power management driver related stuff? (I have no real clue about this stuff, just speculating)

Factory Factory
Mar 19, 2010

This is what
Arcane Velocity was like.
Also, it does indeed look like there will be process scheduling issues in Windows analogous to Hyperthreading inefficiencies, and though Microsoft is on the case, that there will be some time before they are resolved. As such, all benchmarks we see (which I haven't even gotten to yet, personally) will be filling threads inefficiently.

e: Single-threaded performance :ughh:



All turbo and power saving disabled, for the record.

Factory Factory fucked around with this message at 04:34 on Oct 12, 2011

Alereon
Feb 6, 2004

Dehumanize yourself and face to Trumpshed
College Slice
Distressing fact: The DIPS/clock/core (number of Dhrystone integer operations per clock cycle, per core) for Bulldozer is 3.78, which is pretty similar to the upcoming ARM Cortex A15 at 3.5. For comparison, Sandy Bridge scores 9.43.

Canned Sunshine
Nov 20, 2005

CAUTION: POST QUALITY UNDER CONSTRUCTION



You know it's a fairly disappointing release when Anandtech doesn't even have a review up when the NDA expired...

Factory Factory
Mar 19, 2010

This is what
Arcane Velocity was like.
That said, I'm getting to the benchmarks (finally) in Tom's review, and the multicore performance really is pretty good. Even hobbled by whatever inefficiencies may be from the module-not-cores architecture, the FX-8150 is holding up extremely well in content creation apps. Sometimes it's the out-and-out winner between the 2600K, 2500K, and 1100T, sometimes it's in between those, and sometimes...

Well, okay, sometimes it's outperformed by a quad-core Phenom II. Still, my expectations have been minimized sufficiently that this is slightly impressing me.

fart simpson
Jul 2, 2005

DEATH TO AMERICA
:xickos:

Factory Factory posted:

Also, it does indeed look like there will be process scheduling issues in Windows analogous to Hyperthreading inefficiencies, and though Microsoft is on the case, that there will be some time before they are resolved. As such, all benchmarks we see (which I haven't even gotten to yet, personally) will be filling threads inefficiently.

e: Single-threaded performance :ughh:



All turbo and power saving disabled, for the record.

Well, you did pick the single worst performing benchmark. It's not like the single-threaded performance really is as bad as this across all of them.

Alereon
Feb 6, 2004

Dehumanize yourself and face to Trumpshed
College Slice

Factory Factory posted:

That said, I'm getting to the benchmarks (finally) in Tom's review, and the multicore performance really is pretty good. Even hobbled by whatever inefficiencies may be from the module-not-cores architecture, the FX-8150 is holding up extremely well in content creation apps. Sometimes it's the out-and-out winner between the 2600K, 2500K, and 1100T, sometimes it's in between those, and sometimes...

Well, okay, sometimes it's outperformed by a quad-core Phenom II. Still, my expectations have been minimized sufficiently that this is slightly impressing me.
I would consider an 8-core processor that can't quite equal a quad-core to be a pretty serious failure. The HardOCP Cinebench numbers show Bulldozer BARELY beating a Phenom II X6, and losing slightly to the i7 2600K. Things are a bit better for POVRay, but I'd definitely say that multi-threaded performnace is far below expectations. I never really expected per-core performance to be good, but I at least thought it would win pretty handily in heavily multi-threaded integer workloads, and that is definitively not the case. I would have also hoped that per-thread floating point performance would go up over Phenom II, but instead it seems to have dropped, pretty seriously when you consider that Bulldozer has a 200-500Mhz clock speed advantage, depending on how effective Turbo Core is.

Bonus Edit: Anandtech's review has been delayed because Anand Lal Shimp was hospitalized yesterday (he's fine now). They're hoping to have it up "soon", which should hopefully mean tonight.

Alereon fucked around with this message at 05:24 on Oct 12, 2011

Factory Factory
Mar 19, 2010

This is what
Arcane Velocity was like.

MeramJert posted:

Well, you did pick the single worst performing benchmark. It's not like the single-threaded performance really is as bad as this across all of them.

True, but it's still disappointing from a desktop use point of view. That said, it turns out that Windows 8 Dev Preview has all the "treat modules as modules, not cores" code in it, so the FX-8150 actually makes up a little ground in lightly-threaded games on Windows 8 instead of Win 7, plus it saves power.

Alereon posted:

I would consider an 8-core processor that can't quite equal a quad-core to be a pretty serious failure. The HardOCP Cinebench numbers show Bulldozer BARELY beating a Phenom II X6, and losing slightly to the i7 2600K. Things are a bit better for POVRay, but I'd definitely say that multi-threaded performnace is far below expectations. I never really expected per-core performance to be good, but I at least thought it would win pretty handily in heavily multi-threaded integer workloads, and that is definitively not the case. I would have also hoped that per-thread floating point performance would go up over Phenom II, but instead it seems to have dropped, pretty seriously when you consider that Bulldozer has a 200-500Mhz clock speed advantage, depending on how effective Turbo Core is.

On one hand, yes, you are absolutely right. Compared to where Bulldozer should be given the state of technology and its competition, both in-house and out, it's disappointing. That said, there are some usage profiles where, if you want to buy $200-$300 of CPU, Bulldozer can stand and deliver. It's not a total shut-out.

I'm the guy in a prison camp thrilled about being beaten only six days a week instead of daily, here.

Factory Factory fucked around with this message at 05:28 on Oct 12, 2011

Longinus00
Dec 29, 2005
Ur-Quan

Alereon posted:

I would consider an 8-core processor that can't quite equal a quad-core to be a pretty serious failure. The HardOCP Cinebench numbers show Bulldozer BARELY beating a Phenom II X6, and losing slightly to the i7 2600K. Things are a bit better for POVRay, but I'd definitely say that multi-threaded performnace is far below expectations. I never really expected per-core performance to be good, but I at least thought it would win pretty handily in heavily multi-threaded integer workloads, and that is definitively not the case. I would have also hoped that per-thread floating point performance would go up over Phenom II, but instead it seems to have dropped, pretty seriously when you consider that Bulldozer has a 200-500Mhz clock speed advantage, depending on how effective Turbo Core is.

Bonus Edit: Anandtech's review has been delayed because Anand Lal Shimp was hospitalized yesterday (he's fine now). They're hoping to have it up "soon", which should hopefully mean tonight.

Cinebench and POVRay are integer workloads?

freeforumuser
Aug 11, 2007
http://www.anandtech.com/show/4955/the-bulldozer-review-amd-fx8150-tested

Anandtech review is up!

BlackMK4
Aug 23, 2006

wat.
Megamarm
I can barely believe they released this thing.

WhyteRyce
Dec 30, 2001

BlackMK4 posted:

I can barely believe they released this thing.

At some point with a project that has been delayed so long you have to decide to either scrap it or just release it as is and hope to recoup whatever you can.

WhyteRyce fucked around with this message at 05:44 on Oct 12, 2011

Devian666
Aug 19, 2008

Take some advice Chris.

Fun Shoe
Between reading tomshardware and anandtech I'm satisfied that AMD have almost caught up to their main competitor the Phenom II X6 1100T.

WhyteRyce
Dec 30, 2001

So after all this, I wonder what the hell was the point of JFAMD posting all over the place. All he seemed to do was either condescendingly trash and discredit leaked benchmarks as being completely false when they turned out to be pretty close to the mark and post deliberately vague information and statements which ended up being completely wrong (no delay for BD, IPC improvement over and over again).

I doubt if he's ever heard again on any of those enthusiast boards because he seems to have drawn the ire of just about everyone.

movax
Aug 30, 2008

gently caress, I just feel terrible for the AMD engineers. They've had a tough road in developing this chip and they've released knowing full well that it's about to get poo poo upon by everyone. Kind of like sending your kid out onto the field knowing he's about to get his rear end kicked into the ground at worst, and barely manage to keep pace with the other kids at best.

That said, the weapon they can bring to bear on Intel is pricing. I wouldn't have an issue tossing a chip like this into a system for a non-gaming, non-techie family member if the price for the mobo and CPU was right. One thing I like about AMD is how long that socket has lasted, and the relative lower costs of their boards. The Intel PDG is a pretty thick book with exacting specifications on every little thing; the AMD guidebook is a little looser, and the specs are pretty tolerant, so you can shortcut a bit at the artwork stage.

Just remember, the consumer is hosed if AMD ceases to be a viable competitor in the desktop x86 market space.

Alereon
Feb 6, 2004

Dehumanize yourself and face to Trumpshed
College Slice

Longinus00 posted:

Cinebench and POVRay are integer workloads?
I certainly could be wrong, but I thought the actual workload was primarily integer, and some quick Googling seems to verify this.

Edit: Yes I was in fact wrong.

Alereon fucked around with this message at 06:42 on Oct 12, 2011

Adbot
ADBOT LOVES YOU

freeforumuser
Aug 11, 2007




:suicide::ughh::negative::psyduck:

And those aren't even crossing the 60 fps barrier.

  • Locked thread