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
kapinga
Oct 12, 2005

I am not a number

Factory Factory posted:

That's literally the question that was asked and answered in the two posts above yours.

Is the answer answer is sometime "after 1H '13"? Because that's the only way I can read that image to say anything at all about Ivy Bridge-E, if it will ever exist.

Adbot
ADBOT LOVES YOU

Factory Factory
Mar 19, 2010

This is what
Arcane Velocity was like.


Yes. IVB-E has dropped off the roadmap without comment from Intel.

movax
Aug 30, 2008



kapinga posted:

Is the answer answer is sometime "after 1H '13"? Because that's the only way I can read that image to say anything at all about Ivy Bridge-E, if it will ever exist.

1H'13 sounds like a good guess to me; they're really hauling rear end on Haswell as well.

DaNzA
Sep 11, 2001

:D


Grimey Drawer

Yeah that's what made me ask that question actually, because I didn't see it on the graph at all.

And "after 1H '13" means they will still be one generation behind again compare to the mainstream

Factory Factory
Mar 19, 2010

This is what
Arcane Velocity was like.


Since they're basically the cast-off dregs of much more heavily tested and validated server offerings, it does kinda make sense. Remember how Intel took a year longer than any other SSD manufacturer to come out with a SandForce SF-2281 SSD.

Agreed
Dec 30, 2003

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



DaNzA posted:

Yeah that's what made me ask that question actually, because I didn't see it on the graph at all.

And "after 1H '13" means they will still be one generation behind again compare to the mainstream

What's the situation where the "premium" class (guess current nomenclature is -E, so we'll go with that) is a good idea, again? It's got way too much power for desktop use, and it's under-featured rather dramatically for server use, while costing as much as a solid investment in either because of the parts premiums. I look at that stuff as Intel's GTX _70, a way to keep making money on parts that didn't make the cut for fully intended purposes - except with the GTX _70 cards, you're at least getting a discrete product that makes market sense, slotting in closer to price:performance optimality. The -E CPUs from Intel on the other hand just seem like a huge waste of money since depending on what you're using the system for, you can do better in either direction.

Maybe an exception is that with SB-E and its more flexible overclocking (offsets letting you get 2700K levels of overclock without buying the top dollar parts), you could get a computer with way more DDR3 RAM than standard P67/Z68 systems, I guess, for a price that's not quite up into what it costs to build a server parts system that would need pricier ECC RAM. But that's a reeeeally narrow performance category, I don't see the selling point.

Alereon
Feb 6, 2004

Dehumanize yourself and face to Trumpshed

College Slice

-E series CPUs exist solely for niche applications that are heavily multi-threaded or dependent on memory bandwidth, which really isn't a lot of things. That said, the higher-end CPUs have always been extremely TDP limited due to the increased core count and uncore size. Sandy Bridge-E is an 8-core die, but Intel can only leave 6 of them enabled while still fitting into a 125W TDP with useful clockspeeds. With Ivy Bridge able to fit four cores plus graphics into 77W, it seems like a hypothetical Ivy Bridge-E would be able to deliver 8 cores with pretty nice clockspeeds in 125W. Then again, they may just skip it and go directly to Haswell-E with its quad-channel DDR4.

hobbesmaster
Jan 28, 2008



Alereon posted:

-E series CPUs exist solely for niche applications that are heavily multi-threaded or dependent on memory bandwidth, which really isn't a lot of things.

Those are he applications that make the world go round though. Intel probably figures Sandy bridge EN/EP will be fine until haswell so there won't be anything to downgrade to the regular E part. Or something.

Alereon
Feb 6, 2004

Dehumanize yourself and face to Trumpshed

College Slice

Almost nobody runs those applications on their desktop computer though, which is why it's a pretty tiny niche.

hobbesmaster
Jan 28, 2008



Alereon posted:

Almost nobody runs those applications on their desktop computer though, which is why it's a pretty tiny niche.

Which is why I mentioned the EN/EP xeons. They won't hit until around the release of Haswell, right? That'd be around when we would see ivy bridge e; if ever.

Bing the Noize
Dec 21, 2008

by The Finn


please excuse my ignorance, as I haven't built anything since the LGA 775 days -

given I'm going to use it on a DAW, is L2 cache of as much importance for work like multitrack real time audio processing or audio encoding as it was on C2D's and such? specifically, i remember the general consensus at the time is that larger L2 caches were greatly beneficial when it came to multitrack recording with lots of other audio processing involved.

for reference, i was thinking about the i3 2120 and the i5 3570K.

Factory Factory
Mar 19, 2010

This is what
Arcane Velocity was like.


In general, yes, as additional cache lowers the likelihood that the CPU will have to make a latency-intensive request to RAM. L3 cache, as well.

Where's Agreed when you need him? He does real-time audio professionally and is literally oozing with knowledge about this stuff.

Bing the Noize
Dec 21, 2008

by The Finn


That's perfect then. I definitely want to hear what he has to say about this, then.
I might cast out the i3 2120 as an option - I don't have the budget to go much higher than a 3570K, but I would still really love to hear more about the current state of CPUs with real-time audio. I hate going to audio-related boards for advice on this stuff because most of the time people there don't know anything about how computers actually work (or aren't very active).

I'm only just moving past my C2D setup which is really rapidly aging and not keeping up at all. I'm going to be using XP, though, because I'm using an E-MU 1616 PCIe with some other older audio software that works best under XP.

The same machine will be running OS X though, and I'll surely be doing audio work there. It's just that the XP DAW setup is what I'm basing my new build around. Planning on using it with 8GB (so OS X can see the other 4GB) and a Biostar Z77 board.

Bing the Noize fucked around with this message at 08:31 on Jun 30, 2012

Shaocaholica
Oct 29, 2002

Fig. 5E


Alereon posted:

Almost nobody runs those applications on their desktop computer though, which is why it's a pretty tiny niche.

I bet a lot of freelancers would rather have a 1 socket 1x8 core machine than a 2 socket 2x4 core or 2x8 core machine. Not to mention how much more expensive the latter is going to be from limited parts supply and choices.

Hell, even at some of the bigger studios I've worked at that switched to dual socket workstations for artists can probably switch back to single socket+high core count if its cheaper and more energy efficient.

Agreed
Dec 30, 2003

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



ACID POLICE posted:

That's perfect then. I definitely want to hear what he has to say about this, then.
I might cast out the i3 2120 as an option - I don't have the budget to go much higher than a 3570K, but I would still really love to hear more about the current state of CPUs with real-time audio. I hate going to audio-related boards for advice on this stuff because most of the time people there don't know anything about how computers actually work (or aren't very active).

I'm only just moving past my C2D setup which is really rapidly aging and not keeping up at all. I'm going to be using XP, though, because I'm using an E-MU 1616 PCIe with some other older audio software that works best under XP.

The same machine will be running OS X though, and I'll surely be doing audio work there. It's just that the XP DAW setup is what I'm basing my new build around. Planning on using it with 8GB (so OS X can see the other 4GB) and a Biostar Z77 board.

Hey, caught most of this in the general building thread.

You're really handicapped by XP, though - I know it's ungodly expensive but really consider getting a more modern audio interface, both for features and for compatibility, as well as unshackling the performance of your software. You're bottlenecked hardcore by the lack of proper driver support for your old hardware, and XP doesn't handle multiple cores, or SSDs, or any of the cool stuff that made my upgrade from a Q9550-based setup with 8GB of RAM and dual-booting XP 32-bit and Vista 64-bit (back then, I had compatibility issues too - which were thankfully addressed as time went on, leading to me ditching the XP partition entirely after a couple years).

Some specifics - does L2 and L3 cache matter? Heck yeah, for continuous processing you want as little compromising the execution pipeline as possible. You can avoid dropouts at incredibly low latencies with modern processors because they can just plain handle it.

That said, the difference in real-time performance between the i5 and the i7 isn't really commensurate with the price difference; especially since REAPER already excels at multithreading audio, to a degree that I personally think is pretty much unmatched in other DAWs, it was a ground-up feature inclusion and so that part of the audio engine just never had to deal with any compromises or hacks to make it work or all that jazz. But Hyperthreading comes in handy at higher clock rates, giving you a chance to do perfectly acceptable pipeline efficient execution when an i5 might have to waste some cycles. You'll see measurably lower total CPU utilization with an i7 than an i5, just 'cause audio is an area where you do get the benefits of the i7. But if it's a budget wall, nothing at all wrong with the i5 with a -K, a nice cooler, and a stable overclock. Dramatic performance increase over stock - which is already going to feel like lightning compared to a C2D.

Thinking forward you're really going to want to get out of XP. A good audio interface of yesteryear isn't enough reason anymore to keep yourself stuck to it as soon as you can afford something better. But even with that handicap, you'll still notice dramatic performance gains upgrading the system itself to current-gen stuff. This is one of the most powerful runs that Intel's had in... Well, they really just haven't made a mistake since the P4s, and that pays huge dividends to folks like you or me who do audio on a desktop system, not going off the deep end with SB-E but still choosing to use fewer but faster cores as opposed to more but slower cores.

Bing the Noize
Dec 21, 2008

by The Finn


Thanks man, that's exactly what I wanted to know about!
I don't want to clutter up this thread with much non-Intel talk, but for now my plan is to use the E-MU under XP to record and have another partition for OS X with all my Mac audio software, which is much more relevant/up to date/able to make use of modern hardware.

In the near future when I'm making more money again and I find a device that sounds at least as good as the E-MU, I'd love to upgrade, but for now I'll pull the trigger on that 3570K. There's no reason I can't give the E-MU a shot on Windows 7 on a computer that isn't a goddamn dinosaur.

Also, thanks for your comments in the system building thread, both were definitely very useful. I've had pretty good experience with Biostar to date but I am of course open to other options.

Bing the Noize fucked around with this message at 18:54 on Jun 30, 2012

Chuu
Sep 11, 2004



Grimey Drawer

hobbesmaster posted:

Those are he applications that make the world go round though. Intel probably figures Sandy bridge EN/EP will be fine until haswell so there won't be anything to downgrade to the regular E part. Or something.

When you start pricing systems out there's so much overlap between comparable Xeons and the -E parts I assume they only exist so there is some use for the ultra-terrible bins of Xeons.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

Another new socket for the Haswell?!

Factory Factory
Mar 19, 2010

This is what
Arcane Velocity was like.


Yep. The VRM has been moved on-chip, so the power delivery has to be rearranged.

hobbesmaster
Jan 28, 2008



Doesn't seem unreasonable that each tick-tock set would have a new socket.

HalloKitty
Sep 30, 2005

Adjust the bass and let the Alpine blast


hobbesmaster posted:

Doesn't seem unreasonable that each tick-tock set would have a new socket.

Isn't that tock-tick?
(new-shrink)

Toast Museum
Dec 3, 2005

30% Iron Chef


HalloKitty posted:

Isn't that tock-tick?
(new-shrink)

That has always seemed completely backwards to me.

hobbesmaster
Jan 28, 2008



HalloKitty posted:

Isn't that tock-tick?
(new-shrink)

Clocks go tick tock so I will write it that way.

(you're of course right)

Agreed
Dec 30, 2003

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



Toast Museum posted:

That has always seemed completely backwards to me.

I've always found the tick-tock (tock-tick) ((tiock-toick??)) thing to be an incredibly effective mindfuck to the competition as well as a great way to ensure that projects come to fruition internally even if they aren't fully realized (e.g. Ivy Bridge, which I think we can mostly agree is mainly about lithographic improvements with a side order of more integration, not a performance leap or anything close to it, and with some problems that don't matter for the most part but which are indicative of either a rushed development cycle, like PCI-e 3.0 and iffy implementations thereof - or engineering compromises, like the whole "how'sabout we don't use a much more effective heat transfer mechanism this time around, just slap some goopy TIM under that sucker and get it out the door?").

Especially given that AMD had nothing remotely comparable, the specifics of what makes a Tick or a Tock and whether you're really getting the whole benefit proposed by the development cycle structure seems less significant than the simple fact that tick, tock, tick, tock happens, in a way that makes their progress look inexorable compared to their major competitor(s?). Big ol' bags of money don't hurt either.

Shaocaholica
Oct 29, 2002

Fig. 5E


What applications show the biggest gain from SNB-E memory bandwidth and extra cache over vanilla SNB?

Factory Factory
Mar 19, 2010

This is what
Arcane Velocity was like.


Mainly VM farms, scientific simulations, graphics rendering, and high-volume video transcoding and compression/decompression (assuming the algorithm is sufficiently threaded).

Honestly, though, unless you're doing these on an enterprise scale, the difference between dual channel and quad channel on an i7-3960X is nil. A desktop or workstation workload just doesn't need that much RAM bandwidth with an SNB-E i7's enormous cache.

Agreed posted:

(e.g. Ivy Bridge, which I think we can mostly agree is mainly about lithographic improvements with a side order of more integration, not a performance leap or anything close to it...

Man, tell that to HD 4000. It musta didn't hear the plan because it snuck in 40-60% better performance than HD 3000.

Shaocaholica
Oct 29, 2002

Fig. 5E


Factory Factory posted:

Mainly VM farms, scientific simulations, graphics rendering, and high-volume video transcoding and compression/decompression (assuming the algorithm is sufficiently threaded).

Honestly, though, unless you're doing these on an enterprise scale, the difference between dual channel and quad channel on an i7-3960X is nil. A desktop or workstation workload just doesn't need that much RAM bandwidth with an SNB-E i7's enormous cache.

Hmm. Still, it doesn't seem like SNB-E costs all that much more so its not like its a huge waste of money.

mayodreams
Jul 4, 2003


Hello darkness,
my old friend


I've been building dual socket Xeon workstations for a graphics company that I am consulting for. I've found that the EVGA SR2 boards have almost all the features of a desktop (integrated sound, USB3, standard or server ram, etc) that a lot of server and workstation boards don't have. A couple of them (before I got there) are the Sandy Bridge-E variants, but we found the boost in power was better with dual 6 core Xeons.

Yes, the market is small, but these guys have a huge tech budget and only use machines 2-3 years before they replace them because the return on investment is so great.

Agreed
Dec 30, 2003

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



Factory Factory posted:

Man, tell that to HD 4000. It musta didn't hear the plan because it snuck in 40-60% better performance than HD 3000.

And when they can pair that with a processor that isn't top-end in its class, I'll say ooh/ahh about it. As it is, it's basically just a different number that *also* runs additional monitors.

Mobile CPUs with it, has this list grown since may?
i7-3920XM, i7-3820QM and i7-3720QM

Desktop CPUs with it, well, the usual suspects in the high end.

People aren't buying Ivy Bridge because they can play Angry Birds real real good in Chrome or whatever, the only CPUs featuring that impressive numbers improvement are generally paired (even when it comes to large chains etc.) with a graphics card that makes the jump from HD3000 to HD4000 trivial.

I'm not saying it's not something they can mark down as an accomplishment - I just filed it under superior integration and left it there, because for most people it won't ever mean anything at all compared to the HD3000 despite the performance delta.

Factory Factory
Mar 19, 2010

This is what
Arcane Velocity was like.


All mobile Core CPUs have HD 4000, just like SNB all had HD 3000. Even ULV chips have the higher IGP, though it's thermals-limited.

Agreed
Dec 30, 2003

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



God drat it where do you find the TIME to know everything about computers

Well, if it's in all the laptops then that's neato. How's driver support? Was totally crap with the HD3000... And is anything still happening with Optimus, or is nVidia kinda taking their toys and going home since 1. Intel's sort of trying to put them out of business in the mobile sector, and 2. Intel made, like, one pass at updating Optimus drivers, for their end of the bargain, and that's about it...

Factory Factory
Mar 19, 2010

This is what
Arcane Velocity was like.


It's... better? A few months ago, they fixed all the big outstanding HD 3000 bugs (like the refresh rate switching when switching between battery and plug power). The driver team does actual game-specific enhancements. But it's baby stuff compared to AMD or Nvidia.

Also knowing everything about computers is a handy trick of being able to remember practically every sentence written on AnandTech, and also the existence of Wikipedia.

HD 4000's big claim to fame is that it matches Intel CPU cores with a GPU capable of matching a first-gen AMD APU. Where HD 3000 was good in a pinch, HD 4000 is legit okay.

AFAIK, the big thing with Optimus right now is trying to decouple the drivers, which AMD is doing, too. As in, you can install arbitrary versions of Catalyst/Nvidia along with arbitrary versions of Intel HD Graphics, and you don't need to hope and pray that your laptop's OEM does all the necessary proprietary optimizations every time there's a new release. Which would be FANTASTIC news for Sony customers, since Sony releases drivers for maybe four months after a system launches and then halts development.

HalloKitty
Sep 30, 2005

Adjust the bass and let the Alpine blast


Toast Museum posted:

That has always seemed completely backwards to me.

I agree. I don't know why it's that way.

doomisland
Oct 5, 2004



Ke$ha should do a rendition of her song and call it Tock Tick and make it about Intel's product line.

WhyteRyce
Dec 30, 2001



doomisland posted:

Ke$ha should do a rendition of her song and call it Tock Tick and make it about Intel's product line.

Intel blew its budget on will.i.am and that Korean female pop group.

Shaocaholica
Oct 29, 2002

Fig. 5E


They're also bedfellows with Dreamworks:

https://www.youtube.com/watch?v=N1HebuI3ho8

Not sure what the exact agreement is but I'm pretty sure we get a lot of our hardware for pennies on the dollar if at all. Yes, I said 'our'. I worked on this spot :/

Factory Factory
Mar 19, 2010

This is what
Arcane Velocity was like.


Shaocaholica posted:

They're also bedfellows with Dreamworks:

https://www.youtube.com/watch?v=N1HebuI3ho8

Not sure what the exact agreement is but I'm pretty sure we get a lot of our hardware for pennies on the dollar if at all. Yes, I said 'our'. I worked on this spot :/

The group shot in the end only has three smugbrows. I think that's a record low smugbrow density for Dreamworks.

Shaocaholica
Oct 29, 2002

Fig. 5E


Factory Factory posted:

The group shot in the end only has three smugbrows. I think that's a record low smugbrow density for Dreamworks.

That's probably because it was mostly farmed out to another studio. Our animators know better.

German Joey
Dec 18, 2004


WhyteRyce posted:

Intel blew its budget on will.i.am and that Korean female pop group.

wait what? i know about will.i.am, but what's this Korean female pop group?

edit: ahaha oh christ, just goggled this and now I'm laughing my rear end off. at least they didn't give them all blue badges...

Adbot
ADBOT LOVES YOU

FISHMANPET
Mar 3, 2007

Sweet 'N Sour
Can't
Melt
Steel Beams


So I'm trying to figure out the difference between E5-2600 and E5-2400. The 2400 is newer and has a lower TDP compared to a similair 2600, but the 2400 seems to have a slightly lower clockspeed (but I don't know if they're too different for a clock speed comparison to be meaningless). Any thing I can read about this? Is there a reason I should pick 2400 over 2600 (which means I pick a Dell R620/R720 vs a R320/R420).

I'm so confused.

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