Intel Core Ultra 9 285K, Ultra 7 265K and Ultra 5 245K Review Roundup

lopoetve said:
One of the most revolutionary releases of all time. That, sandy bridge for consumers, Athlon 64, the original Pentium MMX 200...
Click to expand...
The Pentium 200 MMX wasn't all that amazing. It wasn't the top model and was identical to the Pentium 200MHz performance wise outside of the handful of applications that supported MMX instructions. The Pentium MMX family was short lived as they were supplanted by the Pentium II's a short time later.
 
Dan_D said:
The Pentium 200 MMX wasn't all that amazing. It wasn't the top model and was identical to the Pentium 200MHz performance wise outside of the handful of applications that supported MMX instructions. The Pentium MMX family was short lived as they were supplanted by the Pentium II's a short time later.
Click to expand...
nintendo cpus (or just cartridge cpu)
 
Dan_D said:
The Pentium 200 MMX wasn't all that amazing. It wasn't the top model and was identical to the Pentium 200MHz performance wise outside of the handful of applications that supported MMX instructions. The Pentium MMX family was short lived as they were supplanted by the Pentium II's a short time later.
Click to expand...
I was mostly thinking the entire top end of the Pentium line at that time - the switch from 90/100/120/133 to the 166/200/233 was significant, at least to me and my friends at the time. Especially since most of us were on 90s and jumped all the way to 200.
 
lopoetve said:
I was mostly thinking the entire top end of the Pentium line at that time - the switch from 90/100/120/133 to the 166/200/233 was significant, at least to me and my friends at the time. Especially since most of us were on 90s and jumped all the way to 200.
Click to expand...
The Pentium line already went to 200MHz prior to the release of the MMX parts. Intel had 133MHz, 166mhz, 150MHz and 200MHz offerings at that point. The desktop lineup for the Pentium MMX series only encompassed 166MHz, 200MHz and a 233MHz part. Mobile offerings had 120MHz, 133MHz, and 150MHz, SKU's. Reviews at the time showed that a 200MHz Pentium and Pentium MMX 200MHz were virtually identical performers in any application that didn't support MMX instructions. At the time almost nothing did unless it was from Microsoft or Adobe.

Intel went all in on the Pentium II lineup and the Pentium MMX was relegated to budget systems pretty quickly before being phased out of OEM machines a very short time later. Retail product cycles at the time could be extremely short. To give an example of this HP even had one Pavilion model that was only on the market for three weeks before being supplanted by the exact same machine. The only difference being that the former had a Pentium II 400MHz processor in it and the latter had a Pentium III 450MHz CPU in it. The incremented the model up by 50 and that was it.
 
Dan_D said:
The Pentium line already went to 200MHz prior to the release of the MMX parts. Intel had 133MHz, 166mhz, 150MHz and 200MHz offerings at that point. The desktop lineup for the Pentium MMX series only encompassed 166MHz, 200MHz and a 233MHz part. Mobile offerings had 120MHz, 133MHz, and 150MHz, SKU's. Reviews at the time showed that a 200MHz Pentium and Pentium MMX 200MHz were virtually identical performers in any application that didn't support MMX instructions. At the time almost nothing did unless it was from Microsoft or Adobe.

Intel went all in on the Pentium II lineup and the Pentium MMX was relegated to budget systems pretty quickly before being phased out of OEM machines a very short time later. Retail product cycles at the time could be extremely short. To give an example of this HP even had one Pavilion model that was only on the market for three weeks before being supplanted by the exact same machine. The only difference being that the former had a Pentium II 400MHz processor in it and the latter had a Pentium III 450MHz CPU in it. The incremented the model up by 50 and that was it.
Click to expand...
That's what I meant - the entire "top end" of those (MMX or not) coming out. The later models there 166/200 were what a few of us were able to jump to, and it was a massive change from the older 90/75/100 that we had. Tagged in MMX without thinking about it (although I did have a demo of some racing game that got some neat effects with MMX).
 
Dan_D said:
Except that's entirely wrong. Intel first integrated the memory controller with Nahalem. It's never changed that until Arrow Lake. Previous CPU's, including the Core and Core 2 families all had memory controllers built into the PCH/Chipset. Not the CPU.
Click to expand...
Yep, I remember buying the i7 920 with triple channel mem back in the day. IIRC that was the first time they moved the IMC on die (Athlon 64 already had it on die) and have now moved it off die again.
 
Removing hyperthreading and the IMC off die seems like a bad idea. Still runs hot, not as efficient as they make it out to be, lower frequencies and lower overclocking headroom. The only benefit seems to be faster encoding speeds, and an NPU, which no one really wants. Expensive new motherboard platform and expensive newer memory types, but why would anyone want to pay a premium for slower overall performance? At least in the past, they made new releases faster, and refreshes to focus on efficiency. They sure missed the ball on this one.
 
El_Capitan said:
Removing hyperthreading and the IMC off die seems like a bad idea. Still runs hot, not as efficient as they make it out to be, lower frequencies and lower overclocking headroom. The only benefit seems to be faster encoding speeds, and an NPU, which no one really wants. Expensive new motherboard platform and expensive newer memory types, but why would anyone want to pay a premium for slower overall performance? At least in the past, they made new releases faster, and refreshes to focus on efficiency. They sure missed the ball on this one.
Click to expand...
There is actually quite a bit of efficiency in there. But, its not at max juice. LIkely due to being made for/based on Laptop design goals.

1731109294524.png
 
chameleoneel said:
There is actually quite a bit of efficiency in there. But, its not at max juice. LIkely due to being made for/based on Laptop design goals.

View attachment 690541
Click to expand...
Yeah, I felt like they wanted to focus mainly at the laptop/mini-pc market. Basically make laptop/mini-pc CPU's, but retrofit them to the desktop market, rather than two vastly separate designs. Kind of like when Microsoft was making their Windows 8 design for their Surface laptop/tablets, and retrofitting it to their desktop users. We all know how that panned out...
 
El_Capitan said:
Removing hyperthreading and the IMC off die seems like a bad idea. Still runs hot, not as efficient as they make it out to be, lower frequencies and lower overclocking headroom. The only benefit seems to be faster encoding speeds, and an NPU, which no one really wants. Expensive new motherboard platform and expensive newer memory types, but why would anyone want to pay a premium for slower overall performance? At least in the past, they made new releases faster, and refreshes to focus on efficiency. They sure missed the ball on this one.
Click to expand...
They removed hyperthreading in place of something different, it’s not like they just went “well let’s yank it”, the chips are using some new fandangled process optimizer that is supposed to just cram jobs where they fit.
On paper it could take one physical core and divide it into 8 if need be while also being core agnostic so it could put parts 9 and 10 onto the NPU and GPU all while the OS is blind to what’s happening.

But the Intel software and drivers are just lacking. It’s not working in reality close to how it does on paper, maybe it’s a RAM or Cache thing?
 
Lakados said:
They removed hyperthreading in place of something different, it’s not like they just went “well let’s yank it”, the chips are using some new fandangled process optimizer that is supposed to just cram jobs where they fit.
On paper it could take one physical core and divide it into 8 if need be while also being core agnostic so it could put parts 9 and 10 onto the NPU and GPU all while the OS is blind to what’s happening.

But the Intel software and drivers are just lacking. It’s not working in reality close to how it does on paper, maybe it’s a RAM or Cache thing?
Click to expand...
Do they not have prototypes and testing? Obviously they had a reason to remove hyperthreading, but something else should have compensated for the lack of it, and I doubt it's software being the issue. Maybe they just thought it would lower power consumption while improving gaming FPS (which some people with 12th -14th gen did), and people would just go gaga over it?

They also slowed the frequency of the ring bus from 5GHz to 3.8GHz, which may have been due to the 13th and 14th gen issues? That definitely contributed to slower performance as well.

We'll have a better picture with the 300 series if those perform better.
 
El_Capitan said:
Removing hyperthreading and the IMC off die seems like a bad idea.
Click to expand...
The only wanted and possible benefit of doing things like that would be cheaper cost, R&D and manufacturing AMD chiplet, Intel tile, etc... hard to judge without knowing how much they achieved to save.

El_Capitan said:
Obviously they had a reason to remove hyperthreading,
Click to expand...
Security issues could be one, but could be e-core being good enough and good thread director for them (better performance on them than on extra logical core) and Intel cpu all have really high core counts now, it feel that decision is perfectly fine at least at first look, when you look at the benchmark where hyperthreading shined the most like blender:
https://www.techpowerup.com/review/intel-core-ultra-7-265k/10.html

The new intel seem to be about the best at it and really impressive, 20 threads 265k beat the 24 threads 9900x by a good amount and the 32 threads 14900k by a little bit, a possible lack of logical thread seem to be the least of their issues here.

El_Capitan said:
They also slowed the frequency of the ring bus from 5GHz to 3.8GHz, which may have been due to the 13th and 14th gen issues?
Click to expand...
Having the memory controller being on a different tile sound like something that could make it harder to go high frequency for the interconnect between different part of the cpu
 
Starting at around 13 minutes, Robert Hallock of Intel, says they know what's wrong. Its a combo of things (bios, firmware, OS, etc). And they are on red alert to fix it. And full details will be made public, soon.

View: https://www.youtube.com/watch?v=P2OHRH7221w

timestamped later in the video. He gives a loose ETA on the fixes, for maybe end of November. And then says they intend to publically explain each part of the problem "line-by-line". He also goes on to insist that Arrow Lake will deliver Raptor Lake gaming performance parity, with less power use.

View: https://www.youtube.com/live/P2OHRH7221w?si=NbQ3ey4ynqwWHOzS&t=1563
 
Last edited:
Lakados said:
They removed hyperthreading in place of something different, it’s not like they just went “well let’s yank it”, the chips are using some new fandangled process optimizer that is supposed to just cram jobs where they fit.
On paper it could take one physical core and divide it into 8 if need be while also being core agnostic so it could put parts 9 and 10 onto the NPU and GPU all while the OS is blind to what’s happening.

But the Intel software and drivers are just lacking. It’s not working in reality close to how it does on paper, maybe it’s a RAM or Cache thing?
Click to expand...
That was Royal Core - it got canned over a year ago (was going to be Jim Keller's next big thing). Basically cores that can divide down (which is a thing and happened on other platforms) rather than fixed thread counts and access.

But it's dead.

https://www.notebookcheck.net/Intel...ed-Beast-Lake-Next-architecture.882771.0.html

https://www.notebookcheck.net/Revol...dded-efficiency-and-performance.695343.0.html

edit: Canned "earlier this year"

El_Capitan said:
Do they not have prototypes and testing? Obviously they had a reason to remove hyperthreading, but something else should have compensated for the lack of it, and I doubt it's software being the issue. Maybe they just thought it would lower power consumption while improving gaming FPS (which some people with 12th -14th gen did), and people would just go gaga over it?

They also slowed the frequency of the ring bus from 5GHz to 3.8GHz, which may have been due to the 13th and 14th gen issues? That definitely contributed to slower performance as well.

We'll have a better picture with the 300 series if those perform better.
Click to expand...
They did - Royal was apparently relatively close to taping out. This looks like bugs and bad design.

Ring Bus - I want to say I saw someone commenting that the new design and power limits were the reason there, and that it also has a major impact.
 
lopoetve said:
One of the most revolutionary releases of all time. That, sandy bridge for consumers, Athlon 64, the original Pentium MMX 200...
Click to expand...
Conroe wasn't all that revolutionary. It was Intel admitting Netburst was a complete failure. The reality was Conroe was Intel resuming what they mistakenly dropped after the Pentium 3 (the very design AMD was benefiting from through the P4 years). Actually, Conroe owes thanks to its predecessor Merom, which was an Intel Israel design. We have them to thank by getting Intel back on track after flopping around like a dying fish with Netburst.

Also, the Pentium Pro was more revolutionary than the Pentium MMX which came after it. MMX itself was overhyped. The Pentium Pro architecture lives on today in each and every processor. Branch prediction, out-of-order execution, speculative execution, register renaming, 64GB of addressable memory via PAE, the list goes on and on. The PPro took everything that was great about the Pentium and just made it that much better. The PPro even placed its L2 cache on the CPU package very close to the CPU die running at or nearly the clock speed of the CPU. In almost every way, the Pentium 2 was a downgrade from the PPro. PPro is a legend. Easily my all-time favorite Intel processor. I'm still holding on to a couple of PPro 200s.
 
Absalom said:
Conroe wasn't all that revolutionary. It was Intel admitting Netburst was a complete failure. The reality was Conroe was Intel resuming what they mistakenly dropped after the Pentium 3 (the very design AMD was benefiting from through the P4 years). Actually, Conroe owes thanks to its predecessor Merom, which was an Intel Israel design. We have them to thank by getting Intel back on track after flopping around like a dying fish with Netburst.

Also, the Pentium Pro was more revolutionary than the Pentium MMX which came after it. MMX itself was overhyped. The Pentium Pro architecture lives on today in each and every processor. Branch prediction, out-of-order execution, speculative execution, register renaming, 64GB of addressable memory via PAE, the list goes on and on. The PPro took everything that was great about the Pentium and just made it that much better. The PPro even placed its L2 cache on the CPU package very close to the CPU die running at or nearly the clock speed of the CPU. In almost every way, the Pentium 2 was a downgrade from the PPro. PPro is a legend. Easily my all-time favorite Intel processor. I'm still holding on to a couple of PPro 200s.
Click to expand...
It was - but going backwards that many years and not only bringing the architecture up to date but making it excel like that? And the fact that the performance was valid and usable well into the Haswell/Broadwell generations? That's revolutionary in my mind. EVERYONE went to snag a Conroe setup. Many of us jumped on Merom laptops! The E8400 and Q6600 chips owned the market for a very long time.

Agreed on the Pro, but that lived in a special world with servers and ultra high end workstations. We plebes didn't really see it unless you were rich or weird. Things have to hit consumer level to really have the same impact. I really freaking wanted one of those things!
 
Comixbooks said:
https://www.forbes.com/sites/antony...cessors-getting-free-major-performance-boost/

Performance Boost incoming...changing the voltage frequency.
Click to expand...
This article is simply rehashing days old info. Intel themselves haven't said anything, since the Robert Hallock interview.

I myself got a little over-excited about Asrock's claims for the new PPM update. But, I had trouble confirming if it was actually installed/working.

Hallock said they will be doing a a "line by line" explanation of the fixes, when ready. I will wait for that.
 
You must log in or register to reply here.
Back
Top