Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

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.
 
  • Post
  • Reply
in a well actually
Jan 26, 2011

dude, you gotta end it on the rhyme

JawnV6 posted:

You could always get your chips from all the other vendors providing transactional memory.

http://dl.acm.org/citation.cfm?id=2593241

Adbot
ADBOT LOVES YOU

Alereon
Feb 6, 2004

Dehumanize yourself and face to Trumpshed
College Slice

JawnV6 posted:

You could always get your chips from all the other vendors providing transactional memory.
I'm not really interested in switching to IBM, thanks though :) Really though, this isn't a minor erratum and Haswell processors have been on the market for 14 months.

r0ck0 posted:

I just bought a 4790k, does this mean they will do a recall or am I just screwed out of this TSX instruction set?
I'm in the same boat, and pretty much yes. We can pray they find a way to fix it in a firmware update, but short of that they'd have to send us new Broadwell processors next year to replace our Haswells. Most likely we'll get a check in the mail a few years from now after Intel settles a class action lawsuit.

Alereon fucked around with this message at 20:06 on Aug 12, 2014

HalloKitty
Sep 30, 2005

Adjust the bass and let the Alpine blast

r0ck0 posted:

I just bought a 4790k, does this mean they will do a recall or am I just screwed out of this TSX instruction set?

My guess is you'll get a BIOS update for your mobo which will hide that functionality.

JawnV6
Jul 4, 2004

So hot ...

Alereon posted:

I'm not really interested in switching to IBM, thanks though :) Really though, this isn't a minor erratum and Haswell processors have been on the market for 14 months.

So you're holding the pre-release validation standard of a completely new software stack's ecosystem to "what can the wild produce in over a year"? This isn't something well understood like TLB shootdowns, it's a fringe case and I think your standards are unreasonable if this is "incompetent".

Number19
May 14, 2003

HOCKEY OWNS
FUCK YEAH


I wonder if that's going to delay the Haswell-E release?

Alereon
Feb 6, 2004

Dehumanize yourself and face to Trumpshed
College Slice

JawnV6 posted:

So you're holding the pre-release validation standard of a completely new software stack's ecosystem to "what can the wild produce in over a year"? This isn't something well understood like TLB shootdowns, it's a fringe case and I think your standards are unreasonable if this is "incompetent".
At the end of the day it's Intel's responsibility to ensure that the products it sells work. Haswell has been on the market for over a year, and just had a fresh re-launch of new SKUs that had TSX added. You don't see me posting about the other documented errata with Intel products because they are minor, needing to disable an ISA feature across all of your processors that has been advertised for over a year because it is plain broken is a huge failure.

movax
Aug 30, 2008

HalloKitty posted:

My guess is you'll get a BIOS update for your mobo which will hide that functionality.

The BIOS loads the microcode blob at every boot; when you update your BIOS, the updated microcode will kill TSX. If you want to live dangerously, don't update your BIOS. OSes can also load microcode patches though, so Microsoft will likely issue a Windows Update that will disable TSX as well.

The verification team's life is probably really lovely now (or leading up to confirming the bug).

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!
I suppose it's good that I held off. Not that hardware lock elision nets big rear end performance improvements, but given various highly threaded apps and games, I take anything as soon the various apps and threading libraries support it.

EoRaptor
Sep 13, 2003



Combat Pretzel posted:

I suppose it's good that I held off. Not that hardware lock elision nets big rear end performance improvements, but given various highly threaded apps and games, I take anything as soon the various apps and threading libraries support it.

Yeah, this looked like a great feature to let multi-threaded applications both run more quickly and run more safely, all with very little work needed outside of compiler and library support.

It'll still happen, but now everybody is going to be much more wary of it and implementation will be massively slowed.

I guess I'm waiting for Broadwell Desktop 2015? I think that's what intel meant when they said it would be fixed in the next Broadwell CPUs.

JawnV6
Jul 4, 2004

So hot ...

Alereon posted:

it is plain broken

Sorry, don't have time to learn your new language

BobHoward
Feb 13, 2012

The only thing white people deserve is a bullet to their empty skull
If it took a year to discover I'm assuming the breakage was real subtle.

My main question is whether there are security implications, as can often be the case with errata of this type.

wargames
Mar 16, 2008

official yospos cat censor

Alereon posted:

:siren:Intel's incompetent QA strikes again:siren:

A critical defect has been discovered in the TSX implementation on Haswell and Broadwell-Y processors that will require it to be disabled in a firmware update. The issue will be resolved in the next Broadwell stepping, but unfortunately there is no way to fix TSX on existing products. This loving sucks as TSX was one of the reasons I waited for a Devil's Canyon CPU. No word yet on what resolution will be for owners of affected products.

This affects my 4690?

Alereon
Feb 6, 2004

Dehumanize yourself and face to Trumpshed
College Slice

JawnV6 posted:

Sorry, don't have time to learn your new language
"Plain broken" wasn't a comment on the bug being easy or difficult to find, but on the fact that the feature got disabled so customers no longer derive any value from it. What matters is the impact of the bug, if it could get a footnote in an errata document I wouldn't care, but they had to disable TSX so it's a huge fuckup.

wargames posted:

This affects my 4690?
Yes. Though really almost nobody has a use for TSX right now, so this only matters in the abstract sense that a part of the product you paid for is broken, and that now it won't become useful in the future. There's always a lag between when new instruction set features are implemented and when they start getting used, and now that lag will be longer.

wargames
Mar 16, 2008

official yospos cat censor

Alereon posted:

"Plain broken" wasn't a comment on the bug being easy or difficult to find, but on the fact that the feature got disabled so customers no longer derive any value from it. What matters is the impact of the bug, if it could get a footnote in an errata document I wouldn't care, but they had to disable TSX so it's a huge fuckup.
Yes. Though really almost nobody has a use for TSX right now, so this only matters in the abstract sense that a part of the product you paid for is broken, and that now it won't become useful in the future. There's always a lag between when new instruction set features are implemented and when they start getting used, and now that lag will be longer.

Why can't our 51st state produce stuff correctly.

edit:
this probably mean emulators won't be using TSX for a while :(

wargames fucked around with this message at 04:42 on Aug 13, 2014

japtor
Oct 28, 2005

Number19 posted:

I wonder if that's going to delay the Haswell-E release?
Doesn't look like it. I was curious about Haswell EP cause there was some PR about it shipping to OEMs a few weeks ago, looks like they're in the same boat as the rest of the Haswell parts:

quote:

Update II: Intel has gotten back to us with some more information about how the TSX erratum affects its upcoming Xeon CPUs.

The launch of Intel's high-volume Haswell-EP processors is rapidly approaching, and the TSX errata apparently won't delay that product launch. Instead, a spokesman for the firm informs us that TSX will be available for software developers to enable "for development purposes" on Haswell-EP, so that their code will be "ready for production" once the higher-end Haswell-EX processors arrive at a later date.

In other words, we expect Haswell-EP to ship on schedule with the TSX erratum still etched into its silicon and TSX instructions disabled via a microcode patch. Those who wish to risk working with TSX in Haswell-EP will have the option to enable it via a firmware menu, but Intel recommends waiting for Haswell-EX before using TSX in production systems.

Since the single-socket, enthusiast-oriented Haswell-E processors are based on the same silicon as the lower-end Xeon EP parts, I'd expect the upcoming Core i7 Extreme CPUs to have TSX disabled in microcode, as well.

Agreed
Dec 30, 2003

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

Big picture horrible, small picture "still pretty much completely satisfied with my 2600K setup, just waiting on it to croak to put the pre-refresh Haswell bits together and that'll probably last another several years" - I know that Processors Are loving Hard but still, that seems a rather vital feature to just screw the pooch on. Wow.

Starting to feel some of that lack of competition going on here.

PerrineClostermann
Dec 15, 2012

by FactsAreUseless

Agreed posted:

Big picture horrible, small picture "still pretty much completely satisfied with my 2600K setup, just waiting on it to croak to put the pre-refresh Haswell bits together and that'll probably last another several years" - I know that Processors Are loving Hard but still, that seems a rather vital feature to just screw the pooch on. Wow.

Starting to feel some of that lack of competition going on here.

Tell me about it. I looked at potentially building a new computer. Turns out, if I wanted better performance, I'd have to drop 800 bucks or something stupid to get more cores and slightly worse single-core performance. Or spend 500 bucks on a flagship GPU. Or 300 on faster, larger RAM. In the end, the only way to upgrade a PC build from 2010 without taking out a loan was...to get a SSD. Now I have no idea what to do. Nothing anywhere seems worth it. We live in weird times.

HalloKitty
Sep 30, 2005

Adjust the bass and let the Alpine blast

PerrineClostermann posted:

Tell me about it. I looked at potentially building a new computer. Turns out, if I wanted better performance, I'd have to drop 800 bucks or something stupid to get more cores and slightly worse single-core performance. Or spend 500 bucks on a flagship GPU. Or 300 on faster, larger RAM. In the end, the only way to upgrade a PC build from 2010 without taking out a loan was...to get a SSD. Now I have no idea what to do. Nothing anywhere seems worth it. We live in weird times.

That's fantastic, though, because you can save money and spend it on something else.

Edit: wait, you said 2010. That means Nehalem, so yes, you can definitely benefit from Devil's Canyon. Although if it was a Gulftown 980X or 990X, I'd probably keep that sexy beast for longer, even if it is outdated.

It's only Sandy (2011) and above that I'd say are truly not worth upgrading. Sandy saved a hell of a lot of power over Nehalem.

vv Oh, overclocked 2600K. Yeah, you have no upgrades necessary. Graphics cards and SSDs until we get anything significant.

HalloKitty fucked around with this message at 10:48 on Aug 13, 2014

PerrineClostermann
Dec 15, 2012

by FactsAreUseless

HalloKitty posted:

That's fantastic, though, because you can save money and spend it on something else.

Edit: wait, you said 2010. That means Nehalem, so yes, you can definitely benefit from Devil's Canyon.

It's only Sandy (2011) and above that I'd say are truly not worth upgrading. Sandy saved a hell of a lot of power over Nehalem.

Er, I've got a 2600k I've OC'd to about 4.5GHz. So I lied, I got my stuff summer of 2011. Sorry about that.

Agreed
Dec 30, 2003

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

PerrineClostermann posted:

Er, I've got a 2600k I've OC'd to about 4.5GHz. So I lied, I got my stuff summer of 2011. Sorry about that.

Same here, June of 2011. Has hosted four top-end graphics cards from nVidia and EVGA (those fuckers get way too much of my money when they release a dumb graphics card argh), and I add a bigger or better SSD or whatever every so often too, and at 4.5GHz the damned thing just keeps going. I even BOUGHT all the parts for a Haswell system because in theory PCI-e 3.0 is cool for me, maybe, in a year or so - and I've had that loving box since last August, and it's still current in terms of performance and features, AND I STILL CAN'T BE ARSED TO BUILD IT because why?

DPC latency isn't a problem for my system, and I don't think the slightly higher DPC latency of the Z87 motherboard I've got is going to be a problem for it when my chip eventually dies. Everything else is a drop-in replacement and it'll be only slightly higher performance for . . .

And the . . . is why I am sad that AMD can't do poo poo. I know that the HUGE factor I'm leaving out here is obviously the tremendous gulf in efficiency between my current system and a truly modern one, but given that the difference amounts to less than like 70W at peak system draw (assuming I could overclock the Haswell processor to performance parity or greater) I find it hard to justify taking the time to tear down a perfectly good rig and build up a new one. Revolutionary for laptops (etc.), great, probably awesome for the small form factor crowd (but maybe not since they're a touch hard to cool!), and totally unnecessary until my current comp literally just stops working.

Really jumped the gun on that system but when it goes together it'll be great, but I am genuinely baffled that my four generation old hardware performs, outside of specific optimizations and special instruction sets that it lacks, roughly on par with a similarly priced system today. I also wonder how Intel is going to deliver on their stated performance targets after Haswell, though at this point I have total faith that they will at least make a seriously, amazingly efficient processor. Less and less is that a perk of node shrinks, more and more the point.

Assepoester
Jul 18, 2004
Probation
Can't post for 11 years!
Melman v2

Star War Sex Parrot posted:

Looks like an NDA just lifted for Broadwell previews:

The Tech Report
AnandTech
Ars Technica
So this means nothing for desktops yet, is what I'm gathering from the articles?

Unless you count the NUC?





Alereon posted:

:siren:Intel's incompetent QA strikes again:siren:

A critical defect has been discovered in the TSX implementation on Haswell and Broadwell-Y processors that will require it to be disabled in a firmware update. The issue will be resolved in the next Broadwell stepping, but unfortunately there is no way to fix TSX on existing products. This loving sucks as TSX was one of the reasons I waited for a Devil's Canyon CPU. No word yet on what resolution will be for owners of affected products.
So don't get a Devil's Canyon until they fix it or drop the price of the broken ones?

evilweasel
Aug 24, 2002

PerrineClostermann posted:

Tell me about it. I looked at potentially building a new computer. Turns out, if I wanted better performance, I'd have to drop 800 bucks or something stupid to get more cores and slightly worse single-core performance. Or spend 500 bucks on a flagship GPU. Or 300 on faster, larger RAM. In the end, the only way to upgrade a PC build from 2010 without taking out a loan was...to get a SSD. Now I have no idea what to do. Nothing anywhere seems worth it. We live in weird times.

Yeah but on the other hand, an SSD is massive for making your computer feel faster.

Daviclond
May 20, 2006

Bad post sighted! Firing.

HalloKitty posted:

That's fantastic, though, because you can save money and spend it on something else.

I hadn't gotten around to overclocking my 2500K for the several years I'd owned it. The other week I had a day off so I just went for it and bumped up the turbo boost speed for 3.7 GHz to 4.5 GHz. It was fine and stable with no additional voltage needed (to be fair the stock voltage my chip came with out the factory was reasonably high to begin with).

Now, yeah, fantastic I save money and don't have to upgrade... except I've owned this system a hell of a long time and I want to build a new one anyway :v:

P.N.T.M.
Jan 14, 2006

tiny dinosaurs
Fun Shoe
People have had a lot of fun with the new Pentiums, maybe build a computer for a relative?

evilweasel
Aug 24, 2002

Daviclond posted:

Now, yeah, fantastic I save money and don't have to upgrade... except I've owned this system a hell of a long time and I want to build a new one anyway :v:

Build a media center/steambox PC for your TV or something, which also has the advantage of being cheaper than a brand new bells and whistles desktop and actually getting you something new for your money.

Pimpmust
Oct 1, 2008

The worst part about having a spanking brand new computer that takes like 10 sec to boot with that sweet SSD is going to work and having to wait 20 minutes for the 3 year old Lenovo laptop there to complete booting.

90% of that is of course all the bloatware corporate stuff, but :negative:

KillHour
Oct 28, 2007


Pimpmust posted:

The worst part about having a spanking brand new computer that takes like 10 sec to boot with that sweet SSD is going to work and having to wait 20 minutes for the 3 year old Lenovo laptop there to complete booting.

90% of that is of course all the bloatware corporate stuff, but :negative:

My corporate laptop is faster than my desktop at everything except games (and it's no slouch at those, either). But that's mostly because my boss told me to email him what laptop I wanted when he hired me.

Daviclond
May 20, 2006

Bad post sighted! Firing.

evilweasel posted:

Build a media center/steambox PC for your TV or something, which also has the advantage of being cheaper than a brand new bells and whistles desktop and actually getting you something new for your money.

I just built a home media server thing that acts as an HTPC and if anything it's prompted the itch, not made it go away :shobon:

I did finally get around to replacing the 6-year-old noisy case fans with quiet Noctuas that have PWM so that's good. My case is full tower sized though and I don't need that much space anymore - especially now that all my storage is on the server - so I'm mainly looking forward to doing a mini-ITX or micro ATX build in a year or two (the new release of something like Skylake would probably prompt me) and getting the case size down to something more reasonable.

The Lord Bude
May 23, 2007

ASK ME ABOUT MY SHITTY, BOUGIE INTERIOR DECORATING ADVICE

Daviclond posted:

I just built a home media server thing that acts as an HTPC and if anything it's prompted the itch, not made it go away :shobon:

I did finally get around to replacing the 6-year-old noisy case fans with quiet Noctuas that have PWM so that's good. My case is full tower sized though and I don't need that much space anymore - especially now that all my storage is on the server - so I'm mainly looking forward to doing a mini-ITX or micro ATX build in a year or two (the new release of something like Skylake would probably prompt me) and getting the case size down to something more reasonable.

I sate my itch by posting in the PC part picking thread.

Palladium
May 8, 2012

Very Good
✔️✔️✔️✔️
I only upgraded from a 4.1GHz to a 4790K running @ 4.2GHz stock (equivalent to at least a 4.8GHZ 2500K) for ~$250 by selling my old Z77 mobo and a i5 2400. Otherwise I wouldn't even bothered.

Agreed
Dec 30, 2003

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

Box of parts, not assembled, over a year now. I even had to back my 2600K off its previous 4.7GHz down to 4.5GHz, but its more efficient execution compared to the 2500K has kept it, like, Yorkfield relevant. I just wish I knew how much of that was a lack of market pressure, and how much of it is "no seriously processors are loving HARD TO MAKE." I mean I can toss some transistors together and make a nice sounding distortion pedal, but things just seem ever so slightly more complicated when the transistor count jumps from less than seven for a fairly "complicated" pedal build vs. billions and billions of really shockingly, my-mind-has-difficulty-comprehending-the-scale, "holy poo poo is that an electron cloud visible??" small non-planar transistors.

MaxxBot
Oct 6, 2003

you could have clapped

you should have clapped!!

Agreed posted:

I just wish I knew how much of that was a lack of market pressure, and how much of it is "no seriously processors are loving HARD TO MAKE."

I think it's a degree of both, and the loving hard to make part applies not just to the actual fabrication but also to the microarchitecture. By that I mean that while Intel is still able to keep cramming more transistors on the chip when they want to, there's also the challenge of making those extra transistors translate into meaningful performance gains. With GPUs and server CPUs things are a bit easier because you can just keep using the MOAR CORES approach at least at the current time, while with desktop CPUs most software doesn't make that approach especially useful. There's also the fact that Amdahl's law puts theoretical limits on that approach in many situations even when the software does catch up to increasing core counts.

When more cores doesn't help the only other option is to increase single threaded performance, either by increasing the clock frequency or increasing IPC. The clock frequency gains slowed to a crawl a decade ago so really the only approach that leaves is the IPC approach. To increase IPC you add more cache, more associativity, more advanced branch prediction, more registers, more execution ports, etc but that stuff is all starting to reach diminishing returns AFAIK, meaning that you don't get the same performance increases per amount of added transistors that you used to be able to. There are limits on exploiting instruction level parallelism meaning that just adding more parallel units to the pipeline isn't particularly useful after a while because you'll never keep them all fed.

That said, if Intel has actual competition in the desktop CPU market and there was more demand I'm sure they'd find a way to provide more than the piddly performance increases they now offer. They just wouldn't be able to offer large performance jumps every generation such as with GPUs for example because

Agreed posted:

processors are loving HARD TO MAKE

BurritoJustice
Oct 9, 2012

The Lord Bude posted:

I sate my itch by posting in the PC part picking thread.

The best way to scratch the itch I've found is to build computers for everyone I know (like 20k AUD worth this last year). In lull periods though I usually do build lists for myself and then convince myself I don't need it :shobon:.

HalloKitty
Sep 30, 2005

Adjust the bass and let the Alpine blast

BurritoJustice posted:

In lull periods though I usually do build lists for myself and then convince myself I don't need it :shobon:.

This is the best. I even remember doing this when I was a kid, design several price points for fun.

Sidesaddle Cavalry
Mar 15, 2013

Oh Boy Desert Map
Motherboard aesthetics are a terrible thing to write words about.



:awesome:

HalloKitty
Sep 30, 2005

Adjust the bass and let the Alpine blast

Sidesaddle Cavalry posted:

Motherboard aesthetics are a terrible thing to write words about.



:awesome:

I wonder if we'll ever get one that has most of the small components on the back of the board. There should be enough clearance between the tray and the board based on the standoff size to hide a lot back there. Maybe it's not done because there's no standard for standoff depth?

Sidesaddle Cavalry
Mar 15, 2013

Oh Boy Desert Map
There are cases that attempt partially open standoff layers so that one can fit in bulky CPU backplates, and often high-end GPU makers say "gently caress the form factor" and let users sort out how to get what they bought into their boxes.

If a case can be made for it (better in practice = more sales?), a new design standard would at least get people to try and push adoption, yes. People complaining all the time about RAM being too tall for their giant air coolers? Design a form factor for RAM where the modules bust through the motherboard and ventilate out both sides.

Don Lapre
Mar 28, 2001

If you're having problems you're either holding the phone wrong or you have tiny girl hands.
Or people can just not buy pointless giant ram modules. Also with the amount of people moving to CLC i think ram clearance is becoming less of an issue.

Sidesaddle Cavalry
Mar 15, 2013

Oh Boy Desert Map
It was just a possible example, but you are right in that actually thinking about form factor is much less effort and money than changing the industry, no sarcasm intended.

Adbot
ADBOT LOVES YOU

Rime
Nov 2, 2011

by Games Forum

Don Lapre posted:

Or people can just not buy pointless giant ram modules. Also with the amount of people moving to CLC i think ram clearance is becoming less of an issue.

It kills me that I can't find VLP ram faster than 1333. Like, what in the actual gently caress. :psyduck:

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply