New Intel CPUs

[RejZoR]

7 minutes ago, aisle9 said:

2/4 works fine for most basic computing. 2/2 is only good for Linux these days.

Even 2c/4t should be gone, really. If quad core can't be the minimum standard 13 years after quad core debut, I don'tknow when. Especially now that 8c/16t is basically a norm for not even high end systems, but regular ones.

[aisle9]

14 minutes ago, RejZoR said:

Even 2c/4t should be gone, really. If quad core can't be the minimum standard 13 years after quad core debut, I don'tknow when. Especially now that 8c/16t is basically a norm for not even high end systems, but regular ones.

Eh, mildly disagree overall. Strongly disagree with the idea that 8C/16T is the norm for regular systems. If you're counting the typical off-the-shelf desktop that you'll find in 90% of home studies in America, you're talking mostly about Pentium and i3 systems. 8C/16T still resides mostly on the high end. I think 4C/8T (or even 4/4) is the modern sweet spot for everyday performance, but 2/4 is still a decent option for ultra-low budget machines. It's 2/2 that just flat doesn't work in any OS this side of Linux anymore.

[Bombastinator]

11 minutes ago, aisle9 said:

Eh, mildly disagree overall. Strongly disagree with the idea that 8C/16T is the norm for regular systems. If you're counting the typical off-the-shelf desktop that you'll find in 90% of home studies in America, you're talking mostly about Pentium and i3 systems. 8C/16T still resides mostly on the high end. I think 4C/8T (or even 4/4) is the modern sweet spot for everyday performance, but 2/4 is still a decent option for ultra-low budget machines. It's 2/2 that just flat doesn't work in any OS this side of Linux anymore.

Points to the fact that most computers are not used primarily for gaming.  Most things done with computers don’t need that much horsepower.  It is a primary reason I haven’t just gone 8/16  yet. I think I will eventually need to if I want to game.  From a user perspective it’s all about what devs write for in the future.  If devs continue to write for 6 threads for gaming a cpu bigger than 4/8 will not be needed.  The upcoming consoles are 14 or 15 thread though.  They may change things.  They’re not out yet though.

[aisle9]

3 minutes ago, Bombastinator said:

Points to the fact that most computers are not used primarily for gaming.  Most things done with computers don’t need that much horsepower.  It is a primary reason I haven’t just gone 8/16  yet. I think I will eventually need to if I want to game.  From a user perspective it’s all about what devs write for in the future.  If devs continue to write for 6 threads for gaming a cpu bigger than 4/8 will not be needed.  The upcoming consoles are 14 or 15 thread though.  They may change things.  They’re not out yet though.

Even then, my 6C/12T 8086K isn't something I feel any need to replace, and I do game. Not to the extent I used to, but enough so that I wake up every day and curse my GTX 960.

 

But yeah, that's the broader point I was getting to. The typical individual who walks into Best Buy, Walmart or Staples looking for a PC is going to walk out with one of the following:

• A slimline system with (most likely) a BGA Pentium
• An mATX tower with a socketed i3
• An AIO with an i3 or i5, possibly a laptop U-SKU, inside

And that's exactly where it's designed to be. Intel specifically designs the LGA Pentium and i3 families to fill the everyday market for buyers who have $400 in their pocket and need something to read email on. I know why the 2/2 Celeron is still a thing: OEMs love them so they can do those "doorbuster" sales with a $200 desktop/monitor package, and shoppers won't realize until after they hook everything up just what a pile of poo they've bought. Intel loves them because they can make money on silicon that would otherwise be scrapped. It's going to take Microsoft saying, "Hey, guys, Windows 10 needs more than two threads. Sorry," for the Celeron to either be overhauled or dropped. And if the Celeron ever does become a 2/4 CPU, what are the over/under odds on how long it takes Intel to revive the Atom name for a 2/2 "ultra-low power" desktop part?

[Lord Vile]

2 hours ago, thechinchinsong said:

I see. If it's cheaper, then I could see a reason for celeron. But at that level, i would be tempted to buy used for even cheaper. At that point, a windows license would be more expensive than the build and I could see something like a cheap thin client or uber budget prebuilt being just as reasonable.

TBH i'd look for a used G4560, you can find them for like £20 plus postage and motherboards for not much more 

[Bombastinator]

15 minutes ago, aisle9 said:

Even then, my 6C/12T 8086K isn't something I feel any need to replace, and I do game. Not to the extent I used to, but enough so that I wake up every day and curse my GTX 960.

 

But yeah, that's the broader point I was getting to. The typical individual who walks into Best Buy, Walmart or Staples looking for a PC is going to walk out with one of the following:

• A slimline system with (most likely) a BGA Pentium
• An mATX tower with a socketed i3
• An AIO with an i3 or i5, possibly a laptop U-SKU, inside

And that's exactly where it's designed to be. Intel specifically designs the LGA Pentium and i3 families to fill the everyday market for buyers who have $400 in their pocket and need something to read email on. I know why the 2/2 Celeron is still a thing: OEMs love them so they can do those "doorbuster" sales with a $200 desktop/monitor package, and shoppers won't realize until after they hook everything up just what a pile of poo they've bought. Intel loves them because they can make money on silicon that would otherwise be scrapped. It's going to take Microsoft saying, "Hey, guys, Windows 10 needs more than two threads. Sorry," for the Celeron to either be overhauled or dropped. And if the Celeron ever does become a 2/4 CPU, what are the over/under odds on how long it takes Intel to revive the Atom name for a 2/2 "ultra-low power" desktop part?

That 6/12 may or may not last any longer than the 4/8. It also might. Gets more likely if you overclock it.  It all depends on what game developers do.  My understanding is intel used to push programming for 4 threads but moved to pushing for 6 threads.  The issue as I understand it is it turned out that 4 threads is still one plus a few, but 6 threads is “many” so thread count may provide massive advantages.  Remember back in the day when 2.4ghz dual cores trounced 5ghz single cores By 30% or more.  Is 8/16 needed for things now?  No.  Could radically change in November.  I personally wouldn’t buy anything under 8/16 myself if I needed it to game in 6 months.  Internet screen streaming is happening though which means a cpu capable of running a game by itself may no longer be needed.  It seems big iron may come back.  

[aisle9]

1 minute ago, Bombastinator said:

That 6/12 may or may not last any longer than the 4/8. It also might. Gets more likely if you overclock it.  It all depends on what game developers do.  My understanding is intel used to push programming for 4 threads but moved to pushing for 6 threads.  The issue as I understand it is it turned out that 4 threads is still one plus a few, but 6 threads is “many” so thread count may provide massive advantages.  Remember back in the day when 2.4ghz dual cores trounced 5ghz single cores By 30% or more.  Is 8/16 needed for things now?  No.  Could radically change in November.  I personally wouldn’t buy anything under 8/16 myself if I needed it to game in 6 months.  

I also remember back in the XP/Vista era as quads were first making their way to consumer hardware, and so much of the internet lost its shit because a $200 E8500 was spanking a $500 Q6600 in just about every game ever released at the time.

 

But yeah, if I were to upgrade, and I won't barring a disaster, it would be to a Ryzen 7. I don't trust my QBX to handle the heat coming off of a Ryzen 9.

[porina]

1 hour ago, TVwazhere said:

Unless these celerons can OC, they're DOA. 

The majority use cases for these, OC is absolutely irrelevant.

 

8 minutes ago, Bombastinator said:

Remember back in the day when 2.4ghz dual cores trounced 5ghz single cores By 30% or more.

I don't. Care to elaborate? 5 GHz single core was never a thing, outside possibly exotic cooling, but my memory doesn't go far back into single core era. Are we in P4 era here?

[Bombastinator]

8 minutes ago, aisle9 said:

I also remember back in the XP/Vista era as quads were first making their way to consumer hardware, and so much of the internet lost its shit because a $200 E8500 was spanking a $500 Q6600 in just about every game ever released at the time.

 

But yeah, if I were to upgrade, and I won't barring a disaster, it would be to a Ryzen 7. I don't trust my QBX to handle the heat coming off of a Ryzen 9.

Yeah.  I’ve still got one of those quads in my basement as emergency backup.  They had lower single thread than the duos and everything was still totally single core at the time.  2 core helped with single core but 4 didn’t unless multiple apps were up.  That old 4 core still runs stuff though. Not very well of course.  The Ipc sucks for one thing, and there’s more than one thing.  

[Bombastinator]

4 minutes ago, porina said:

The majority use cases for these, OC is absolutely irrelevant.

 

I don't. Care to elaborate? 5 GHz single core was never a thing, outside possibly exotic cooling, but my memory doesn't go far back into single core era. Are we in P4 era here?

There were some celerons that could do it with big cooling.  Moderately early in the introduction of core2duo. Those tests basically killed one core chips for desktops.  Core2duo May have been p3.  I don’t remember. 

[porina]

Just now, Bombastinator said:

There were some celerons that could do it with big cooling.  Moderately early in the introduction of core2duo. Those tests basically killed one core chips for desktops.  Core2duo May have been p3.  I don’t remember. 

In Core 2 era having 2 cores was already commonplace. From memory, major generations went P4, Core, Core 2, Core i. Even the later P4s were dual core so Core 2 was about the 3rd Intel generation supporting dual core.

[Bombastinator]

25 minutes ago, porina said:

In Core 2 era having 2 cores was already commonplace. From memory, major generations went P4, Core, Core 2, Core i. Even the later P4s were dual core so Core 2 was about the 3rd Intel generation supporting dual core.

Yah.  I had a p3 coppermine once long ago.  I don’t remember if it was single or dual though

[soldier_ph]

So to summarize:

 

AMD: Releases 3600XT, 3800XT and 3900XT Chips with a 100MHz clock speed Improvement to celebrate Zen 2s 1 year success and is going strong with the goal of releasing Zen 3 this fall/Holiday Season with an advanced and improved 7nm Node. 5nm is coming next year persumably.

 

 

Intel: Releases the i9-10850k with a 100Mhz fewer clock speed than the 10900K, is stuck on the 14nm Node since 2014 and has absolutely squeeze every bit of performance out of it that they can. It is long overdue for them to move over to 10nm and at the point they can actually produce enough 10nm Chips it will already be considered an old process Node like 14nm is now. 

 

 

AMD just has to gain as much Market Share in the Mobile Segment as they did in the Deskop and Server Segment if only OEMs wouldn't make bad Laptops with the Renoir Chips but I'm sure they will catch up to speed in the next two years in the Mobile Segment as well.

[Kisai]

3 hours ago, RejZoR said:

Even 2c/4t should be gone, really. If quad core can't be the minimum standard 13 years after quad core debut, I don'tknow when. Especially now that 8c/16t is basically a norm for not even high end systems, but regular ones.

4c only became standard last year. Up to that point, even Intel was still selling 2c U/Y parts and OEM's like HP and Dell were still selling them as stable platform parts.

 

Despite them taking substantial amounts of time to do anything, even with the same SSD's in the high end machine.

 

Like really you need to draw the line somewhere between 4 cores being the minimum for the kind of sloppy "html5" application messes that are now passed off, 8GB of ram due to the sloppy messes inefficient "fork" style of multiprocessing rather than multithreading (which html5 software has been doing for the last 10 years,) and how nearly everything is built with the /O2 or /O3 flag which wastes memory instead of the /Os flag, which saves space.

 

Like the only reason we've even needed faster PC's to run office and windows over the last 10 years is entirely due to switching to worse development models that favor resolution-independence, sloppy frameworks and sandboxing as a solution to security rather than maybe just linting code and not releasing code that has an encyclopedia worth of warnings in it like typical open source software tends to have, I can't imagine how many bugs ship in closed source software.

 

I know this sounds like sky-is-falling, but it really does feel like we've being going backwards.

https://www.extremetech.com/internet/311877-chrome-might-not-eat-all-your-ram-after-adopting-this-windows-feature

Quote

Chromium Edge rolled out to almost all Windows 10 computers in the recent May 2020 update. According to Microsoft, this update also implemented a new memory management feature in Edge known as SegmentHeap. In the latest version of Windows, developers can opt into SegmentHeap to lower the RAM usage of a program. Microsoft says it already added support to the new Edge browser, and it has seen a 27 percent drop in the browser’s memory footprint. 

Sounds good right? Fast forward to today.

https://bugs.chromium.org/p/chromium/issues/detail?id=1102281

Quote

You seriously need to reconsider the plan to postpone enabling this - the vast majority of PC users are not going to notice the CPU cost, but are being impacted in overall system performance because of the memory requirements of Chrome.
 

 

Well, now the CPU requirements of Chromium-based browsers increased by 5-10%, 13% increase in energy use, for a 75% reduction in memory usage.

 

Let's not forget that all the Spectre/Meltdown and similar bugs in CPU's have cost us up to 20% of existing CPU performance unless you have a Zen 2 or 10th gen Intel CPU.

[5x5]

59 minutes ago, Pascal... said:

So to summarize:

 

AMD: Releases 3600XT, 3800XT and 3900XT Chips with a 100MHz clock speed Improvement to celebrate Zen 2s 1 year success and is going strong with the goal of releasing Zen 3 this fall/Holiday Season with an advanced and improved 7nm Node. 5nm is coming next year persumably.

 

 

Intel: Releases the i9-10850k with a 100Mhz fewer clock speed than the 10900K, is stuck on the 14nm Node since 2014 and has absolutely squeeze every bit of performance out of it that they can. It is long overdue for them to move over to 10nm and at the point they can actually produce enough 10nm Chips it will already be considered an old process Node like 14nm is now. 

 

 

AMD just now has to gain as much Market Share in the Mobile Segment as they did in the Deskop and Server Segment if only OEMs wouldn't make bad Laptops with the Renoir Chips but I'm sure they will catch up to speed in the next two years in the Mobile Segment as well.

Basically yes. Unless intel use Pitchford magic, they'll be losing ground till 2022

[aisle9]

1 hour ago, porina said:

In Core 2 era having 2 cores was already commonplace. From memory, major generations went P4, Core, Core 2, Core i. Even the later P4s were dual core so Core 2 was about the 3rd Intel generation supporting dual core.

As much as Intel would like you to, please don't forget the wonderful Pentium D

 

The later P4 CPUs, some of the Socket 478s and all of the LGA 775s, had hyper-threading, but none of them were true dual-cores.

[dfsgsfa]

18 hours ago, zeusthemoose said:

core i9 will be called the i9-10850k and is essentially a 10900k with 100 MHz lower

there is only one way to read this, faulty /badly binned chips failed to become 10900

[porina]

9 minutes ago, aisle9 said:

As much as Intel would like you to, please don't forget the wonderful Pentium D

Pentium D is a dual core P4, so I'd include it in that family.

[Sauron]

3 hours ago, aisle9 said:

I picked a Skylake one up several years ago with a G3900 (?) for $70 on clearance at Walmart, dropped in a sub-$150 i5-6400, then flipped the thing as a home office/media center PC for $400.

Your ability to scam people isn't really an argument in favor of this thing's existence.

[Sauron]

2 hours ago, Bombastinator said:

Core2duo May have been p3.  I don’t remember. 

Pentium 3s came out 7 years before the C2 line. The Core 2 series is what brought Intel back from half a decade of essentially trash products, the Pentium 3 being pretty much the last good one before the C2.

24 minutes ago, dfsgsfa said:

there is only one way to read this, faulty /badly binned chips failed to become 10900

Either that or Intel knows that at flagship price they aren't going to sell a whole lot of 10900k chips.

[RejZoR]

@Kisai

Quad core becoming a standard last year (2019)? Excjuzme, what?! Quad core CPU's are not anything exotic since Q6600. 13 years ago. Sure those were the beginnings, but there were so many totally not enthusiast level quad core processors ages ago. I know tons of casuals who had AMD Athlon X4 CPU's. Those were nothing super exotic. A lot of people had Q8400 and Q9400 models. These were probably around same era. Hell, even Bulldozer AMD FX-8350 were no special exotic. Technically a quad core with dual execution units, but whatever. Those were like 100-120€ CPU's. 8 years ago! Hell, there were Core i5's that were quad cores several years ago, even before Skylake. Whatever their codename was. Bottom Sandy Bridge i5 models? Still 9 years ago. Yeah, quad core should be a bottom standard and dual cores with HT shouldn't exist. I mean for god sake even my tablet has a quad core. It's Atom, sure, but a frigging tablet has it...

[dfsgsfa]

1 minute ago, Sauron said:

Either that or Intel knows that at flagship price they aren't going to sell a whole lot of 10900k chips.

also proven again , intel never drops its price tag

[aisle9]

34 minutes ago, Sauron said:

Your ability to scam people isn't really an argument in favor of this thing's existence.

Knowingly selling a defective or dying item, or something that is not as you claim it to be = scam

Disclosing the exact condition and configuration of a system, putting a price on it and waiting until someone meets that price =/= scam. It = capitalism.

 

If anything, at the time, that $400 was an excellent price for an i5-based system. Not my fault that Walmart put it on fire sale clearance and I just happened to make a good profit on the back end because of it.

 

 

[Bombastinator]

7 minutes ago, aisle9 said:

Knowingly selling a defective or dying item, or something that is not as you claim it to be = scam

Disclosing the exact condition and configuration of a system, putting a price on it and waiting until someone meets that price =/= scam. It = capitalism.

 

If anything, at the time, that $400 was an excellent price for an i5-based system. Not my fault that Walmart put it on fire sale clearance and I just happened to make a good profit on the back end because of it.

 

 

True.  Cheating someone is not always a scam. It’s semantics though.  The effect is the same.  The question is similar to the difference between “stolen” and “stolen fair and square”

[Sauron]

28 minutes ago, aisle9 said:

Knowingly selling a defective or dying item, or something that is not as you claim it to be = scam

Disclosing the exact condition and configuration of a system, putting a price on it and waiting until someone meets that price =/= scam. It = capitalism.

Sounds like capitalism is a scam if you put it that way, considering you did basically no work to flip someone else's product.

 

But hey, technically what you did isn't illegal so I guess technically it doesn't meet the definition of a scam. Good job for meeting that extremely low bar I suppose.

 

The object being broken is irrelevant, you're still lying about the value. The value of what you sold was $70+$150 + minor assembly work = $250 at best.

 

So again, the possibility of using it in a few years to get someone to overpay for something else is not really a good reason for Intel to sell it.