What specs do you think a 20th Anniversary Celeron(s) will have?

[epsilon84]

Those were the days.



The analogy would be if they made an unlocked 6C/12T running at 3GHz which could then run at 4.5GHz for ~$70. So its... unlikely...

I like the idea of a anniversary edition dual core non-HT CFL @ 4.5GHz. Could be useful for certain things. And not much else.

Useful for a non gaming, non productivity PC. So basically, a YouTube box

I actually had a G3258 overclocked to 4.5GHz for a while, it was $50 and good fun, though from 2016 onwards it was practically obsolete because games had moved beyond 2 threads.

 

[crashtech]

IF they make a 20th anniv. Celeron, they'll probably just unlock one and slot it in between the top Celeron and bottom Pentium.

I'm not sure if Celeron is a brand Intel really cares to draw much attention to, though. Pentium was different because it used to be the name of their fastest chips.

 

[Insert_Nickname]

Useful for a non gaming, non productivity PC. So basically, a YouTube box

I actually had a G3258 overclocked to 4.5GHz for a while, it was $50 and good fun, though from 2016 onwards it was practically obsolete because games had moved beyond 2 threads.

The G3258 was a fun little OC toy. Coupled with fast memory (requiring a Z-series chipset) it was actually useful for running older poorly multi-threaded games and unoptimised software on the cheap.

Though I'd never use such a thing in an actual production environment.

 

[cbn]

The G3258 was a fun little OC toy. Coupled with fast memory (requiring a Z-series chipset) it was actually useful for running older poorly multi-threaded games and unoptimised software on the cheap.

Yes, I actually found my G3258 would run DDR3 2133 (haven't tried DDR3 2400 yet) when I updated the BIOS on my MSI Z97 U3 plus.

Would be interesting to once again test with the 33% increase in bandwidth (previously I was running DDR3 1600 with it).

 

[whm1974]

Useful for a non gaming, non productivity PC. So basically, a YouTube box

I actually had a G3258 overclocked to 4.5GHz for a while, it was $50 and good fun, though from 2016 onwards it was practically obsolete because games had moved beyond 2 threads.

Pretty much the only use for 2c/2t processors in basic PCs these days. If I was a x86 CPU manufacturer I wouldn't bother with producing anything less with then 4 threads.

 

[Topweasel]

What about a rather low-clocked / no-turbo 4C/4T CPU? That would blow some minds. Maybe blow Intel's product stack up too. I'm thinking something like AMD Sempron 3850 AM1 CPU, a 1.3Ghz quad-core. Only, 2.0Ghz for the Anniv. Celeron.

Edit: And to make it true to form, it WOULD have "crippled" cache memory. (Aka, Celeron-esque L3 cache amounts, rather than i5-esque.) To ensure that these low-budget chips wouldn't be used by OEMs in place of real i5s. (Well, maybe every OEM except Packard Bell and whitebox builders.)

If they want it in the vein of the original Celeron it would have no L3 cache, be multiplier locked, and be clocked really low.

 

[LTC8K6]

When you could run the G3258 unlocked on a non-Z board, it was a blast to play with. Not so much after Intel clamped down.

 

[SPBHM]

Old Celerons could be overclocked to match Intel's fastest CPU. In honor of that, I propose an unlocked 2GHz 18 core Celeron.

well, for the 300A and newer, but if they want to honor the OG Celeron they better remove the l2/l3 caches!

 

[cbn]

well, for the 300A and newer, but if they want to honor the OG Celeron they better remove the l2/l3 caches!

In one sense the current Celeron (and actually Pentium as well) are different from the higher end chips in that they don't have the ability to cache via M.2 NVMe Optane. (This is listed in the various processor specs as Optane memory support).

But maybe in keeping with the theme of 300A, Intel ends up keeping that M.2 Optane cache differentiation intact (ie, Optane memory cache via M.2 NVMe still ends up not being suported on the 20th Anniversary Celeron(s)).....but instead they add some "300A style on processor cache" in the form of Optane placed very close to the 20th Anniversary Celeron processor. This would also boost 4K QD1 read tremendously compared to the 4K QD1 read we see with the M.2 NVMe Optane. (re: latency would drop on the Optane from something like 10us to ~500ns if the PCIe bus interface was switched to one that use DDR or something similar to DDR)

 

[NTMBK]

In one sense the current Celeron (and actually Pentium as well) are different from the higher end chips in that they don't have the ability to cache via M.2 NVMe Optane. (This is listed in the various processor specs as Optane memory support).

But maybe in keeping with the theme of 300A, Intel ends up keeping that M.2 Optane cache differentiation intact (ie, Optane memory cache via M.2 NVMe still ends up not being suported on the 20th Anniversary Celeron(s)).....but instead they add some "300A style on processor cache" in the form of Optane placed very close to the 20th Anniversary Celeron processor. This would also boost 4K QD1 read tremendously compared to the 4K QD1 read we see with the M.2 NVMe Optane. (re: latency would drop on the Optane from something like 10us to ~500ns if the PCIe bus interface was switched to one that use DDR or something similar to DDR)

Yeah but nobody cares about Optane.

 

[Insert_Nickname]

Yeah but nobody cares about Optane.

I care about Optane. But not for caching...

If you have some spare funds available, try one of the 32GB variety as a boot drive. You'll be surprised, even with Windows image compression.

 

[NTMBK]

I care about Optane. But not for caching...

If you have some spare funds available, try one of the 32GB variety as a boot drive. You'll be surprised, even with Windows image compression.

 

[TheELF]

In one sense the current Celeron (and actually Pentium as well) are different from the higher end chips in that they don't have the ability to cache via M.2 NVMe Optane. (This is listed in the various processor specs as Optane memory support).

I'm sure that as soon as early adopters wouldn't be pissed off anymore intel will quietly unlock it in the drivers just as they did with QSV in haswell celerons.

 

[Markfw]

Yeah but nobody cares about Optane.

I agree. My 960EVO's boot win 10 in 7 seconds. And load everything instantly, at least I can't time them they are so fast.

 

[Insert_Nickname]

I agree. My 960EVO's boot win 10 in 7 seconds. And load everything instantly, at least I can't time them they are so fast.

That's just it. A 32GB Optane drive performs exactly the same, but is considerably cheaper then a 960PRO/EVO. Its difficult to find really high performance SSD at lower capacities.

For a system that pretty much only needs a drive to run the OS, they are a perfect fit. Outside that particular use case, Optane -is- pretty meh...

 

[Markfw]

That's just it. A 32GB Optane drive performs exactly the same, but is considerably cheaper then a 960PRO/EVO. Its difficult to find really high performance SSD at lower capacities.

For a system that pretty much only needs a drive to run the OS, they are a perfect fit. Outside that particular use case, Optane -is- pretty meh...

I use my boot SD for everything but data, all installed programs, and it was $123. Worth every dime.

 

[dullard]

I use my boot SD for everything but data, all installed programs, and it was $123. Worth every dime.

The point is that a simple Optane memory drive is half that cost and is even faster than your 960 EVO in many uses. Lower price and faster speed is something that most of us should want.

Just because your use case doesn't require anything faster at the moment does not mean that the rest of us don't need faster drives. I have many files (mostly Excel and PDFs) that I use that bring my SSD down to a standstill and would love a faster way to open them. Heck, just booting GIMP is an exercise in frustration waiting for my SSD to open all the files. Note: I don't have Optane yet as I haven't yet purchased my next computer that would be compatible.

Optane certainly isn't for everyone (yet). But the quote above that "nobody cares about Optane" seems to be lacking imagination of where Optane is going. An on-CPU Optane memory chip would be a game changer for many use cases and would add very little to the cost of the CPU. http://www.tomshardware.com/news/intel-emib-interconnect-fpga-chiplet,35316.html

Or memory-intensive uses suddenly become financially viable. Think about self-driving cars that need ~1 GB of memory per second of driving. At current DDR4 prices, that is nearly $10/second. The simple 32 GB Optane memory would take that down to $1.84/second while the Optane 900p would be $1.25/second. Do you really want the car to be using your far slower and 1/85th the endurance 960 EVO for its life and death decisions? I could go on and on with other use business use cases (servers, big data, simulations, etc) that need massive amounts of memory and would love to be 10x cheaper.

 

[dullard]

That said, I don't think Celeron is going the Optane route. Intel just released their cheapest Coffee-Lake motherboard chipsets (H310) and those are the ones WITHOUT Optane support.

 

[Insert_Nickname]

The point is that a simple Optane memory drive is half that cost and is even faster than your 960 EVO in many uses. Lower price and faster speed is something that most of us should want.

The 4K random read performance is off the chart. From storagereview:

Looking at its overall performance, we found the Intel 800p to be a bit uneven at times; however, in certain tests such as 4K read and VDI boot, the latency of the Optane SSDs were dramatically lower than the best-in-class Samsung 960 PRO consumer NVMe M.2 drive. In random performance, both 800P capacities showed a solid peak read performance of 354,00 IOPS, with a sustained latency significantly lower than the 960 PRO throughout the test. In writes, however, the top performing 800P capacity (118GB) performed significantly slower than the 960 PRO, peaking at 147,500 IOPS with 824μs latency.

 

[cbn]

Regarding the performance of Optane I own both a 58GB 800p SSD and late last week I also picked up a 16GB Optane.....one thing I noticed about both (16GB used for a Linux Installation) is that they excel in certain situations where there is low amounts of system memory.

For example while doing internet browsing I noticed that paging out (when RAM is exhausted) actually works better than expected. Also when I paged out ( Windows RAM cache depleted) I notice applications still load fairly quick.

This in one reason I am particularly interested in seeing what a "Celeron 300A" style 20th Anniversary Celeron could do.

 

[Topweasel]

The point is that a simple Optane memory drive is half that cost and is even faster than your 960 EVO in many uses. Lower price and faster speed is something that most of us should want.

Just because your use case doesn't require anything faster at the moment does not mean that the rest of us don't need faster drives. I have many files (mostly Excel and PDFs) that I use that bring my SSD down to a standstill and would love a faster way to open them. Heck, just booting GIMP is an exercise in frustration waiting for my SSD to open all the files. Note: I don't have Optane yet as I haven't yet purchased my next computer that would be compatible.

Optane certainly isn't for everyone (yet). But the quote above that "nobody cares about Optane" seems to be lacking imagination of where Optane is going. An on-CPU Optane memory chip would be a game changer for many use cases and would add very little to the cost of the CPU. http://www.tomshardware.com/news/intel-emib-interconnect-fpga-chiplet,35316.html

Or memory-intensive uses suddenly become financially viable. Think about self-driving cars that need ~1 GB of memory per second of driving. At current DDR4 prices, that is nearly $10/second. The simple 32 GB Optane memory would take that down to $1.84/second while the Optane 900p would be $1.25/second. Do you really want the car to be using your far slower and 1/85th the endurance 960 EVO for its life and death decisions? I could go on and on with other use business use cases (servers, big data, simulations, etc) that need massive amounts of memory and would love to be 10x cheaper.

You are talking in Circles, One that people should get it as an OS drive, then turn around and state that it's useful for things that might cause a SATA SSD to choke up. Aren't those things you wouldn't have on an OS drive?

NVME drives are already overkill. Performance that really means nothing in the world of Desktop/Laptop even professional workloads. Ones that it would help require a second drive or a fast network connection (10Gbe) to robust NAS or device with another NVME drive. In that sense Optane might be a decent scratch drive, but then you start running into possible capacity issues when dealing with high bit rate 4k. Personally I haven't had a 32GB OS or Scratch drive since my first Raptors. What you lose in outright performance compared to Optane with even a SATA SSD is made up easily on storage even if you don't need it. Optane on the desktop is a poor idea, tricked by looking at server performance numbers and trying to apply them to desktop workloads, while reworking your system design process to find a solution that has it make sense. Optane has exactly one solution that makes any sense on desktops and that is the 32GB version as a cache drive for systems with a single spindle disk. Even then the requirements to use it edge into the enthusiast equipment where getting a Sata SSD or smaller NVME drive make more sense than relying completely on the chipsets caching to make the drive seem anywhere near as quick as an SSD boot drive using a Optane drive.

Optane as it is right now, is a proof of concept drive, to get enthusiasts to hear about it's "awesomeness" and buy the memory chips they are making as they try to sell it to the Server world (where it is better served).

 

[dullard]

Optane as it is right now, is a proof of concept drive

I think we are talking past each other. Look at the underlined portion in your post vs the underlined portion in my post:

But the quote above that "nobody cares about Optane" seems to be lacking imagination of where Optane is going.

Yes, Optane RIGHT NOW is not for everyone. But my whole post was about where it will be, not where it is now. I'm not limiting my discussion to just 32 GB OS drives (and therefore much of what you posted), because that isn't where Optane is going. That was just an opening volley to get the Optane product launched. But a 32 GB cache drive is not the future. And this whole thread is about what may come, not about what is available now.

You are talking in Circles, One that people should get it as an OS drive, then turn around and state that it's useful for things that might cause a SATA SSD to choke up. Aren't those things you wouldn't have on an OS drive?

I don't understand your logic here. Optane has three main uses (a) as a cache, (b) as a fast drive, and (c) as a cheaper version of memory. Why must we be limited to discussing just one?

 

[whm1974]

I think we are talking past each other. Look at the underlined portion in your post vs the underlined portion in my post:

Yes, Optane RIGHT NOW is not for everyone. But my whole post was about where it will be, not where it is now. I'm not limiting my discussion to just 32 GB OS drives (and therefore much of what you posted), because that isn't where Optane is going. That was just an opening volley to get the Optane product launched. But a 32 GB cache drive is not the future. And this whole thread is about what may come, not about what is available now.

I don't understand your logic here. Optane has three main uses (a) as a cache, (b) as a fast drive, and (c) as a cheaper version of memory. Why must we be limited to discussing just one?

(b) Once storage space increases to usable level will be the most useful for the majority of users. But that will take awhile to reach.

 

[Topweasel]

I think we are talking past each other. Look at the underlined portion in your post vs the underlined portion in my post:

Yes, Optane RIGHT NOW is not for everyone. But my whole post was about where it will be, not where it is now. I'm not limiting my discussion to just 32 GB OS drives (and therefore much of what you posted), because that isn't where Optane is going. That was just an opening volley to get the Optane product launched. But a 32 GB cache drive is not the future. And this whole thread is about what may come, not about what is available now.

I don't understand your logic here. Optane has three main uses (a) as a cache, (b) as a fast drive, and (c) as a cheaper version of memory. Why must we be limited to discussing just one?

No you were talking in circles. I get what you want to get across, but Optane is a waste as a "fast drive" and isn't feasible as a memory solution for desktop users. People don't need to purchase a 32GB cache drive to "sample" Optane as a highly constrained OS drive. They aren't saving money by doing so versus a larger drive because the management effort that would go into keeping that drive usable isn't worth it (have you looked at the installed size of Windows, without patches, even if you turn Hibernate and the swap file off?). It will be lovely 5 years down the road when Optane Memory caches become a thing on desktops (if it ever is) but you are dreaming if you think Optane is a product anyone should be interested in now and while I know that not purchasing Optane would likely seal its doom before it has a chance to hit it's prime, but suggesting that people get it "because" when it has no real viable use case to those who have the ability to use it is disingenuous.

Going from well it's most cost effective because it's cheaper than a 128-256GB SSD, to bouncing into how much better it for a bunch of specific tasks, is talking in circles, because the only real advantage it has over NVME is in having a 32GB drive for those that only need that. But again the only realistic use for those drives is as a spindle drive user to use as cache.

This isn't some future proofing discussion. A Optane drive isn't like adding more cores, or a big video card so you don't have to worry in a couple years. An Optane drive purchased now, will have all the same performance, the same restrictions, the same nearly useless for Desktop users characteristics no matter how useful the tech becomes in future years. You can't be future looking about a device that will never help you in the future.

 

[cbn]

Optane is a waste as a "fast drive" and isn't feasible as a memory solution for desktop users.

I think it (NVMe Optane) can work as memory for desktop users, but the application cannot be very compute intensive. (Thinking paging out during web browsing rather than paging out during Blender Rendering)