Intel is reversing course and bringing hyper-threading back to its server chips

[DragonSlayer101]

Forward-looking: In a major reversal from its earlier decision to ditch hyper-threading across its entire product portfolio, Intel has officially confirmed that simultaneous multi-threading will return to its data center processors with the Xeon 8 "Coral Rapids" family, scheduled for release in 2028. However, the company did not confirm whether the technology will also return to its client CPUs.

Speaking to investors and analysts at Intel's Q2 2026 earnings call on Thursday, CEO Lip-Bu Tan said that the company is working on several strategies to regain its CPU market share across the retail and enterprise segments, including bringing back simultaneous multi-threading with Coral Rapids in 2028. He added that Intel is also focusing on improving single-threaded performance to regain competitiveness in key market segments.

Tan confirmed the return of hyper-threading while answering a question from Morgan Stanley analyst Joe Moore, who asked if the company had any concrete plans to regain lost market share from AMD and Arm in the next five years. Tan did not directly address Intel's rivalry with AMD and downplayed the competition with Arm, saying that the British chip design firm is a "great partner" for Intel, and that the two companies have a strong working relationship.

After making processors with hyper-threading for more than two decades, Intel ditched the technology with the launch of Arrow Lake-S in 2024. However, the company had already reduced its use of hyper-threading over the previous processor generations, starting with the Alder Lake family in 2021. Alder Lake, which was Intel's first desktop processor family to feature a hybrid core architecture, incorporated hyper-threading only in its performance cores while the efficiency cores implemented only a single hardware thread each.

On the server side, the company's current-generation Xeon 6 Granite Rapids processors retain hyper-threaded performance cores, offering up to 128 physical cores and 256 threads. However, the Xeon 6+ Clearwater Forest family ditches hyper-threading entirely, featuring up to 288 single-threaded E-cores built on Intel's 18A process node. The upcoming Xeon 7 Diamond Rapids lineup is expected to offer up to 192 performance cores without simultaneous multi-threading.

Hyper-threading, which is Intel's proprietary implementation of simultaneous multi-threading, was originally introduced in the early 2000s in single-core Xeon and Pentium 4 processors, promising improved performance in multi-threaded workloads. But with the steady increase in the number of physical cores in x86 CPUs, Intel ditched hyper-threading in the belief that it was largely a relic of the past and not nearly as important as it was twenty years ago.

Permalink to story:

Intel is reversing course and bringing hyper-threading back to its server chips

 

[Puiu]

They really need to achieve the highest clock speeds and IPC they can in order to compete with AMD while using the least amount of power. (in both servers and desktop)

If Zen6 rumours are to believed, Intel will need around 40-50% better gaming performance to get close to AMD (270k being the baseline).

 

[Endymio]

They really need to achieve the highest clock speeds and IPC they can in order to compete with AMD while using the least amount of power. (in both servers and desktop)

The last thing their server chips need are higher clock speeds. Intel's largest server chips are still only 288 cores, and your average server is running 10 to 100 times as many processes. Additional cores are a more efficient way to boost performance in terms of power consumption.

IPC in general terms is helpful-- but the traditional manner of increasing that is a longer pipeline, which means higher energy consumption.

 

[Inthenstus]

The last thing their server chips need are higher clock speeds. Intel's largest server chips are still only 288 cores, and your average server is running 10 to 100 times as many processes. Additional cores are a more efficient way to boost performance in terms of power consumption.

IPC in general terms is helpful-- but the traditional manner of increasing that is a longer pipeline, which means higher energy consumption.

But… he wasn’t talking about servers… I get what you’re saying but that isn’t what they said.

 

[Endymio]

But… he wasn’t talking about servers…

But he literally specified desktops and servers.

 

[PierreJG]

Man, HT is old. I can remember that in the early 2000s I bought my first own gaming pc. Took me almost a year, as I had to buy 1 part per month. They had the 3ghz Northwood and then the Prescott. The difference was that the Northwood had only a 512MB level 2 cache, and the Prescott had a 1MB Level 2 cache. I know there were some with a 2MB level 2 cache, but it was never released here. Back then, the 512MB ran faster in games than the 1MB Prescott. Might have been that 1MB cache memory was a little ahead of its time.

I am still trying to find a good point in the E and P cores. There are benefits, but for an AMD CPU that does the same and is faster in gaming, you pay less here. I do not have a new Intel, but it would be nice to see Intel bring back HT.

 

[Theinsanegamer]

Man, HT is old. I can remember that in the early 2000s I bought my first own gaming pc. Took me almost a year, as I had to buy 1 part per month. They had the 3ghz Northwood and then the Prescott. The difference was that the Northwood had only a 512MB level 2 cache, and the Prescott had a 1MB Level 2 cache. I know there were some with a 2MB level 2 cache, but it was never released here. Back then, the 512MB ran faster in games than the 1MB Prescott. Might have been that 1MB cache memory was a little ahead of its time.

I am still trying to find a good point in the E and P cores. There are benefits, but for an AMD CPU that does the same and is faster in gaming, you pay less here. I do not have a new Intel, but it would be nice to see Intel bring back HT.

The cache is not why it was slower. Prescott had a longer 31 stage pipeline, compared to northwood's 20 stage. The cache was insufficient to overcome the latency at the same clockspeed, and prescott was unable to clock significantly higher then northwood due to power and thermal limitations.

Also, we DID get the 2MB CPUs. Those were Cedar Mill and the extreme edition. You can buy them on ebay right now. Now, nobody BOUGHT them, because core 2 was a thing and absolutely slaughtered the pentium 4 across the board. There was also Gallatin, which had a 2MB L3 cache alongside the 512MB L2. These were fantastic, I had one in my childhood PC, but they were rare and expensive, the L3 cache was neat but situational. For Sins of a solar empire and Supreme commander, it was a significant bump.

The E cores are great for productivity. AMD loses there, no ifs and or buts about it. In gaming, only the x3d are faster, the normal zen 5 tie or lose to Raptor/Arrow lake. They also allow much lower idle power use, very noticeable on mobile, AMD has nothing that can touch the likes of Lunar lake.

 

[Puiu]

But he literally specified desktops and servers.

They absolutely do need higher clocks speeds. It's dumb to think otherwise. Not all tasks can be serialised, you still need powerful cores. You can make 200VMs on a single CPU, but a single bespoke task that has just a single thread will require a fast core.

 

[noblody]

IPC in general terms is helpful-- but the traditional manner of increasing that is a longer pipeline, which means higher energy consumption.

Increasing the IPC is done via wider cores (more operations in parallel) and shorter pipelines (less clocks spent per instruction). The longer pipeline is usually a way of achieving higher operating frequencies.

 

[Endymio]

They absolutely do need higher clocks speeds. It's dumb to think otherwise. Not all tasks can be serialised.

You fail to understand that a server processor is generally running up to 100 tasks on every core, and thus, long before it can finish an "unserializable" task, it has been forced to perform hundreds of highly expensive context switches, and each individual switch can cost up to 10,000 clocks directly, and 10x or more indirectly. A heavily-loaded server CPU can spend 50% or more its time in context switches, and recovering from them.

Now, a 10% increase in clock speeds means a 10% performance boost. But 10% more cores not only gives you that 10% boost, but it also reduces context switching on other cores, for a superlinear performance gain that desktop cpus generally don't see.

If you achieve higher clocks through a smaller process node at the same or lower voltage they're essentially free: why not take them? But today, they generally require longer pipelines or other means which reduce power efficiency, which is far more important for a server chip than desktop.

Increasing the IPC is done via wider cores (more operations in parallel) and shorter pipelines (less clocks spent per instruction). The longer pipeline is usually a way of achieving higher operating frequencies.

Absolutely correct; I was still thinking of clock speeds when I typed my response.

 

[Puiu]

You fail to understand that a server processor is generally running up to 100 tasks on every core, and thus, long before it can finish an "unserializable" task, it has been forced to perform hundreds of highly expensive context switches, and each individual switch can cost up to 10,000 clocks directly, and 10x or more indirectly. A heavily-loaded server CPU can spend 50% or more its time in context switches, and recovering from them.

Now, a 10% increase in clock speeds means a 10% performance boost. But 10% more cores not only gives you that 10% boost, but it also reduces context switching on other cores, for a superlinear performance gain that desktop cpus generally don't see.

If you achieve higher clocks through a smaller process node at the same or lower voltage they're essentially free: why not take them? But today, they generally require longer pipelines or other means which reduce power efficiency, which is far more important for a server chip than desktop.


Absolutely correct; I was still thinking of clock speeds when I typed my response.

You clearly are not tech savvy enough for these kind of arguments. You have no idea wtf you are talking about dude. You have MANY tasks that require a powerful single core.

You can have 10000 CPU cores, if you are limited by what a single core can do then it's useless. All such tasks will take too long, you can only run multiple of them which does not solve the main issue where you need the individual results fast.

Next time stick to lying about nazis and defending pedo trump instead of trying to explain how CPUs work

 

[Endymio]

You clearly are not tech savvy enough for these kind of arguments. You have no idea wtf you are talking about dude. You have MANY tasks that require a powerful single core.

Stop. Just stop this nonsense. My claim is trivially easy to prove. How do chip designers build server CPUs to handle those higher loads?

Intel:
Most powerful desktop CPU: Core i9-14900KS: 24 cores, 6.2 GHZ
Most powerful server CPU: Xeon 6980P, 128 cores, 3.2 GHZ

AMD:
Most powerful desktop CPU: Ryzen 9950X: 16 cores, 5.7 GHZ
Most powerful server CPU: EPYC 9965: 192 cores, 3.7 GHZ

Do you need AI assistance and a few days hard thinking to see the pattern here? Many more cores, but much lower clocks.

The same pattern is true for AWS Graviton and other server CPUs. None have the clock rates of far smaller and cheaper desktop CPUs because core counts matter far more than clock speeds. Period.

Next time stick to lying about nazis and defending pedo trump instead of trying to explain how CPUs work

More thread polluting from the proud resident of the child trafficking capital of the Northern Hemisphere: Romania.

 

[BadThad]

The cache is not why it was slower. Prescott had a longer 31 stage pipeline, compared to northwood's 20 stage. The cache was insufficient to overcome the latency at the same clockspeed, and prescott was unable to clock significantly higher then northwood due to power and thermal limitations.

Also, we DID get the 2MB CPUs. Those were Cedar Mill and the extreme edition. You can buy them on ebay right now. Now, nobody BOUGHT them, because core 2 was a thing and absolutely slaughtered the pentium 4 across the board. There was also Gallatin, which had a 2MB L3 cache alongside the 512MB L2. These were fantastic, I had one in my childhood PC, but they were rare and expensive, the L3 cache was neat but situational. For Sins of a solar empire and Supreme commander, it was a significant bump.

The E cores are great for productivity. AMD loses there, no ifs and or buts about it. In gaming, only the x3d are faster, the normal zen 5 tie or lose to Raptor/Arrow lake. They also allow much lower idle power use, very noticeable on mobile, AMD has nothing that can touch the likes of Lunar lake.

Having lived through this and owned all of those CPU's....this is all fact. GREAT POST!

The Extreme Edition was a riot! The CPU fan NEVER stopped no matter how good of a cooler you had. I couldn't wring an extra 100MHz out of mine without it crashing either.

 

[JamesBlond]

Stop. Just stop this nonsense. My claim is trivially easy to prove. How do chip designers build server CPUs to handle those higher loads?

Intel:
Most powerful desktop CPU: Core i9-14900KS: 24 cores, 6.2 GHZ
Most powerful server CPU: Xeon 6980P, 128 cores, 3.2 GHZ

AMD:
Most powerful desktop CPU: Ryzen 9950X: 16 cores, 5.7 GHZ
Most powerful server CPU: EPYC 9965: 192 cores, 3.7 GHZ

Do you need AI assistance and a few days hard thinking to see the pattern here? Many more cores, but much lower clocks.

The same pattern is true for AWS Graviton and other server CPUs. None have the clock rates of far smaller and cheaper desktop CPUs because core counts matter far more than clock speeds. Period.


More thread polluting from the proud resident of the child trafficking capital of the Northern Hemisphere: Romania.

Ukraine is the traffiking Kingdom

 

[Puiu]

Stop. Just stop this nonsense. My claim is trivially easy to prove. How do chip designers build server CPUs to handle those higher loads?

Intel:
Most powerful desktop CPU: Core i9-14900KS: 24 cores, 6.2 GHZ
Most powerful server CPU: Xeon 6980P, 128 cores, 3.2 GHZ

AMD:
Most powerful desktop CPU: Ryzen 9950X: 16 cores, 5.7 GHZ
Most powerful server CPU: EPYC 9965: 192 cores, 3.7 GHZ

Do you need AI assistance and a few days hard thinking to see the pattern here? Many more cores, but much lower clocks.

The same pattern is true for AWS Graviton and other server CPUs. None have the clock rates of far smaller and cheaper desktop CPUs because core counts matter far more than clock speeds. Period.


More thread polluting from the proud resident of the child trafficking capital of the Northern Hemisphere: Romania.

WTF is this BS? You think that giving some specs changes anything? Pfff hahahahaha. I knew you weren't tech savvy, but this is just embarrassing.

I can do that too:
AMD EPYC™ 4564P - 16 cores, up to 5.7GHzz

You also don't understand the reason why servers don't have such high clock speeds as with desktop chips.

"How do chip designers build server CPUs to handle those higher loads?"
You don't understand the workloads at all. It's absolutely bonkers for me to talk about single threaded tasks and you to keep saying that more cores will solve the problem.

Here you go, actual workload explanation:

AI Innovators Adopt NVIDIA Vera — Why Max Single-Threaded CPU at Scale Matters

NVIDIA Vera exemplifies a new class of CPU, architected for the era of agents and being adopted by AI innovators including Perplexity; NVIDIA CPU roadmap continues with the NVIDIA Rosa CPU and its Rigel core.

blogs.nvidia.com

This is just one of many workloads that need both many cores and maximum single threaded performance.

And next gen AMD CPUs focus heavily on clock speeds with Venice-X toping out at 5.15GHz compared to existing 3.7GHz on the 96core CPUs (full zen6 cores). But hey, you just copy paste random stuff.

In the end all you are doing is showing to the world that you are confidently wrong again.

 

[Endymio]

WTF is this BS? You think that giving some specs changes anything?

Do they "change" anything? No; they simply demonstrate I am correct. Server CPUs prioritize core counts over clock speeds.

You don't understand the workloads at all. It's absolutely bonkers for me to talk about single threaded tasks

Servers don't allocate entire cores to single-threaded tasks to run to completion. Are you sure you even understand what a server is? They typically are running thousands -- sometimes tens of thousands -- of processes, and a 100+ core CPU can be task-switching as often as one million times a second.

Here you go, actual explanation:

AI Innovators Adopt NVIDIA Vera — Why Max Single-Threaded CPU at Scale Matters

NVIDIA Vera exemplifies a new class of CPU, architected for the era of agents and being adopted by AI innovators including Perplexity; NVIDIA CPU roadmap continues with the NVIDIA Rosa CPU and its Rigel core.

blogs.nvidia.com

My god, this is embarrassing. Did you even read your own link? First and most importantly, Vera Max is built for AI agents, and is not intended as a general purpose server CPU. As the article painstakingly explains, it is designed for one specific scenario (running AI agents launched by the GPU) and not for maximum overall CPU performance, but to optimize GPU usage:

"...For the AI factory, the utilization of GPU is the most valuable resource in the data center."

These agentic workloads tend to run a far smaller number of processes -- typically 1-10 per CPU core, rather than the 100+per core of a traditional server.

Finally, I'll note the your own link specifically makes my point for me:

"...While the world has fast CPUs for PCs and workstations, data center CPUs have been evolving in directions away from single-threaded performance. The advent of the cloud has pushed CPU makers to build higher core-count CPUs..."'

This is from YOUR OWN LINK. So the question is: why do you keep arguing absurd positions? I doubt you're even convincing yourself, much less anyone else.

 

[Puiu]

Do they "change" anything? No; they simply demonstrate I am correct. Server CPUs prioritize core counts over clock speeds.


Servers don't allocate entire cores to single-threaded tasks to run to completion. Are you sure you even understand what a server is? They typically are running thousands -- sometimes tens of thousands -- of processes, and a 100+ core CPU can be task-switching as often as one million times a second.


My god, this is embarrassing. Did you even read your own link? First and most importantly, Vera Max is built for AI agents, and is not intended as a general purpose server CPU. As the article painstakingly explains, it is designed for one specific scenario (running AI agents launched by the GPU) and not for maximum overall CPU performance, but to optimize GPU usage:

"...For the AI factory, the utilization of GPU is the most valuable resource in the data center."

These agentic workloads tend to run a far smaller number of processes -- typically 1-10 per CPU core, rather than the 100+per core of a traditional server.

Finally, I'll note the your own link specifically makes my point for me:

"...While the world has fast CPUs for PCs and workstations, data center CPUs have been evolving in directions away from single-threaded performance. The advent of the cloud has pushed CPU makers to build higher core-count CPUs..."'

This is from YOUR OWN LINK. So the question is: why do you keep arguing absurd positions? I doubt you're even convincing yourself, much less anyone else.

This is just embarrassing. You clearly have no idea wtf you are talking about when it comes to server workloads. You have NEVER bought a server in the cloud or used one to understand how things work.

Any traditional database REQUIRES high per core performance. Database transactions are inherently sequential due to ACID compliance which means that you need to finish task A to start task B and this requires high per core performance. Not to mention the fact that database licenses are usually done PER CORE which makes it inherently expensive to use a lot of small cores. So you have to max out each core you pay for.

And have you ever seen how the stock market works? Go look at High-Frequency Trading and what server requirements it has.

"Vera Max is built for AI agents, and is not intended as a general purpose server CPU" - because that's just what Nvidia needs, and that's also why I gave the AMD example of how CPUs are moving forward. But hey, that's too much for you to understand. You think AMD won't be running agentic AI on their CPUs?

" and not for maximum overall CPU performance, but to optimize GPU usage:" - well DUH. Everything is connected. It's why they need maximum single threaded performance. How is this so hard to understand for you? It's just basic stuff.

"The advent of the cloud has pushed CPU makers to build higher core-count CPUs" - it's talking about how it evolved up until now. Tell me, what is the core count of Venice-X? Is it going up? I'll answer because you are too lazy: NO.

About the "-X" server CPUs from AMD:
"This design is purposely built to maximize per-core performance and cache size for high-performance computing (HPC) and technical workloads rather than scaling out raw multi-core density."

Do you think they are making these CPUs because there's no demand for them? Obviously not.

 

[Endymio]

This is just embarrassing. You clearly have no idea wtf you are talking about when it comes to server workloads. You have NEVER bought a server in the cloud or used one to understand how things work.

Saying this nonsense doesn't make it true.

Any traditional database REQUIRES high per core performance.

Then why does every database server CPU in the world have low clock speeds but high core counts?

Database transactions are inherently sequential due to ACID compliance

The fastest databases in the world perform tens of thousands of ACID-compliant transactions per second PER CORE -- and do so using low clocked, high-core count CPUs.

...which means that you need to finish task A to start task B and this requires high per core performance.

You really don't understand the argument at all here, do you? I'll try again, though it's apparent you have no actual interest in facts. If a sequential task is running alone on a dedicated core, then the only way to increase performance is to complete that task faster, which means higher single-core performance. But when you're running 100 or sometimes 1000+ tasks simultaneously per core, then you achieve far greater performance gains by increasing core counts, giving each core less work to perform.

And have you ever seen how the stock market works? Go look at High-Frequency Trading and what server requirements it has.

This is just embarrassing. First of all, HST isn't "how the stock market works". The NYSE, for instance, runs on AWS, using Graviton5 CPUS running at a measly 3.3 GHZ

How The New York Stock Exchange built its real-time market data platform on AWS | Amazon Web Services

This blog discusses how The New York Stock Exchange Group expanded its cloud-based market data product offerings by launching NYSE Cloud Streaming. NYSE Cloud Streaming is a real-time market data streaming solution that lands high-quality market data into AWS with sub-hundred milliseconds...

aws.amazon.com

"The advent of the cloud has pushed CPU makers to build higher core-count CPUs" - it's talking about how it evolved up until now. Tell me, what is the core count of Venice-X? Is it going up? I'll answer because you are too lazy: NO.

Which part of "Venice-X is not a general purpose server CPU" did you fail to understand?

"This design is purposely built to maximize per-core performance and cache size for high-performance computing (HPC) and technical workloads

Do you honestly not understand that HPC and "technical workloads" using GPU-powered AI models are not traditional server workloads?

Just stop. You're digging your hole far deeper. I suggest you take a deep breath and stop allowing your emotions to overwhelm your critical thinking abilities.

 

[Puiu]

Saying this nonsense doesn't make it true.

Then why does every database server CPU in the world have low clock speeds but high core counts?


The fastest databases in the world perform tens of thousands of ACID-compliant transactions per second PER CORE -- and do so using low clocked, high-core count CPUs.

You really don't understand the argument at all here, do you? I'll try again, though it's apparent you have no actual interest in facts. If a sequential task is running alone on a dedicated core, then the only way to increase performance is to complete that task faster, which means higher single-core performance. But when you're running 100 or sometimes 1000+ tasks simultaneously per core, then you achieve far greater performance gains by increasing core counts, giving each core less work to perform.


This is just embarrassing. First of all, HST isn't "how the stock market works". The NYSE, for instance, runs on AWS, using Graviton5 CPUS running at a measly 3.3 GHZ

How The New York Stock Exchange built its real-time market data platform on AWS | Amazon Web Services

This blog discusses how The New York Stock Exchange Group expanded its cloud-based market data product offerings by launching NYSE Cloud Streaming. NYSE Cloud Streaming is a real-time market data streaming solution that lands high-quality market data into AWS with sub-hundred milliseconds...

aws.amazon.com

Which part of "Venice-X is not a general purpose server CPU" did you fail to understand?


Do you honestly not understand that HPC and "technical workloads" using GPU-powered AI models are not traditional server workloads?

Just stop. You're digging your hole far deeper. I suggest you take a deep breath and stop allowing your emotions to overwhelm your critical thinking abilities.

"Then why does every database server CPU in the world have low clock speeds but high core counts?" - this alone makes it so cringe to read your BS

" The fastest databases in the world perform tens of thousands of ACID-compliant transactions per second PER CORE -- and do so using low clocked, high-core count CPUs. " - factually untrue. You are looking at tens of thousands per cluster, not per core. bBut hey, I don't expect you to know that. ACID transactions care deeply about latency (how fast an individual transaction can commit). High core counts and low clock speeds are sub-optimal hardware: this is a known fact.

" If a sequential task is running alone on a dedicated core, then the only way to increase performance is to complete that task faster, which means higher single-core performance." - there is no need to write further. This alone proves my point. You are not running 1000 core heavy tasks per core.

"The NYSE, for instance, runs on AWS" - of all the wacky insane things you could have wrote... no dude, the New York Stock Exchange do not run on AWS or any other public cloud. Educate yourself. They use proprietary servers in a datacenter called "Liquidity Center" located in New Jersey. Only non-critical non-latency sensitive things run on AWS which is why they use cheaper CPUs there. Here's a fact that we know (and I do know this because my brother worked for years as a chief software engineer for the london stock exchange): they use processors optimized for high clock speeds rather than massive core counts for these type of servers.

"Do you honestly not understand that HPC and "technical workloads" using GPU-powered AI models are not traditional server workloads?" - has absolutely nothing to do with what we were talking about and you seem to very confused about the workloads and what these server chips are used for.

I have no idea where you are getting your info... but this has been very eye opening. You are literally just inventing things that have nothing to do with reality. Tech is not for you. What are you even doing here dude? Spreading lies? Getting angry when people correct you with facts?

 

[Endymio]

High core counts and low clock speeds are sub-optimal hardware: this is a known fact.

You don't need to automatically wrong on every issue. Why not try a change? The highest-performing databases in the world run on high core count, low clock speed CPUs.

AWS Graviton: high core count, low clock speed.
Intel Xeon: high core count, low clock speed.
AMD Epyc: high core count, low clock speed.

As the link you gave yourself verifies, server CPUs have long prioritized cores over clocks.

"The NYSE, for instance, runs on AWS" - of all the wacky insane things you could have wrote... no dude, the New York Stock Exchange do not run on AWS or any other public cloud.

Good grief. READ the link I gave. The market data portions of the NYSE run on AWS:

"How The New York Stock Exchange built its real-time market data platform on AWS ... . In 2021, the NYSE Group markets migrated all of their non-real-time data offerings to AWS and shared the key business drivers and results of the migration in an AWS blog the following year....."

Only non-critical non-latency sensitive things run on AWS which is why they use cheaper CPUs there ... they use processors optimized for high clock speeds rather than massive core counts for these type of servers.

LOL, what? The private portions of the NYSE run Linux on HP Proliants -- using high core count, low clock speed Xeon and Epyc CPUs. All of them clocked considerably slower than even the cheapest of desktop CPUs. As does every other high performance database in the world. Let me repeat that: they use lower-clocked CPUs.

Do you think the NYSE has some super-secret underground Arctic laboratory, with cutting-edge silicon fabs producing special CPUs no one knows about? Are you literally being paid to publicly embarrass yourself?

 

[Puiu]

You don't need to automatically wrong on every issue. Why not try a change? The highest-performing databases in the world run on high core count, low clock speed CPUs.

AWS Graviton: high core count, low clock speed.
Intel Xeon: high core count, low clock speed.
AMD Epyc: high core count, low clock speed.

As the link you gave yourself verifies, server CPUs have long prioritized cores over clocks.


Good grief. READ the link I gave. The market data portions of the NYSE run on AWS:

"How The New York Stock Exchange built its real-time market data platform on AWS ... . In 2021, the NYSE Group markets migrated all of their non-real-time data offerings to AWS and shared the key business drivers and results of the migration in an AWS blog the following year....."



LOL, what? The private portions of the NYSE run Linux on HP Proliants -- using high core count, low clock speed Xeon and Epyc CPUs. All of them clocked considerably slower than even the cheapest of desktop CPUs. As does every other high performance database in the world. Let me repeat that: they use lower-clocked CPUs.

Do you think the NYSE has some super-secret underground Arctic laboratory, with cutting-edge silicon fabs producing special CPUs no one knows about? Are you literally being paid to publicly embarrass yourself?

This is just embarrassing seeing you making claims with zero knowledge.

It seems "low clock speeds" for you are coming from a weird comparison you keep making with desktop CPUs. Why? It makes zero sense, especially after I gave you an example of how high click speeds can go with server CPUs that have similar core counts.

"The market data portions of the NYSE run on AWS:" - the transactions for NYSE ALL RUN OF HIGH PERFORMANCE CPUs with high clocks and low core counts. But you fail to understand that this is relative to server hardware. This is a known fact. FYI you cannot know which model of CPU the servers use for their proprietary servers so stop pretending like you do.

The blog you linked clearly tells you that it's not what I'm talking about, but hey, YOU CANNOT READ:

"How the NYSE cut download times and built a cloud-native platform for our historical data products"​

Can you not read "download times" and "historical data products"? You are literally saying the exact same thing I said about their use AWS and you seem to think that these things are related to the high frequency transactions which they are not.

"All of them clocked considerably slower than even the cheapest of desktop CPUs. " - congrats on proving to the world that you do not know anything about servers and wtf you are talking about. It's like I'm talking to a child who refuses to accept that his toy isn't real. Why TF do you keep comparing the clock speeds with desktop CPUs? It's just dumb.

"Are you literally being paid to publicly embarrass yourself?" - you clearly are. why?

 

[Endymio]

It seems "low clock speeds" for you are coming from a weird comparison you keep making with desktop CPUs.

Is this a joke? In the early days of computing, there were no "desktop" or "server" CPUs. There were just CPUs, period. As time progressed, chip designers naturally asked themselves how could they increase performance? And what did they find?

For desktop CPUs, performance was best increased by increasing clock rates.
For server CPUs, performance was best increased by increased core counts.

This is why every server CPU today has 4, 8, or or even 16X as many cores as a desktop CPU -- but much lower clock speeds. I've proven this repeatedly with real-world examples.

Honestly, why do you keep arguing against such basic, self-evident real world facts? Every chip designer in the world disagrees with you. Please: stop the nonsense:

This is a known fact. FYI you cannot know which model of CPU the [NYSE] servers use for their proprietary servers so stop pretending like you do.

Tell us again your absurd fantasy that the NYSE is secretly building their own custom CPUs in hidden underground laboratories.

Now, to once again correct you:

"Underneath the impressive trading floors of the NYSE, a powerful system of servers and processors hums away — processing millions of quotes and trades every second ...Instead of expensive, high-end machines, the NYSE turned to HP ProLiant servers — over 200 four-way HP ProLiant DL585s and 400 ProLiant BL 685c blade servers, all running RHEL...."

How NYSE moved to Linux and Reduced infra cod.

In the fast-paced world of global finance, the New York Stock Exchange (NYSE) stands as an iconic symbol of capitalism and financial power…

overflowbyte.medium.com

"The NYSE and its subsidiaries conduct trading in the United States and five European countries..... Instead of running on proprietary hardware, the exchange runs on 200 four-way HP ProLiant DL585 servers and 400 ProLiant BL 685c blade servers...."

New York Stock Exchange Runs Trades On Red Hat Linux | InformationWeek

The conversion to Linux followed the acquisition of the Euronext exchange in 2007, and the open source operating system is now powering the NYSE's mission-critical trading systems.

www.informationweek.com

Want to guess what CPUs HP ProLiant servers support? (hint: they're not made in underground Arctic laboratories)

 

[Puiu]

Is this a joke? In the early days of computing, there were no "desktop" or "server" CPUs. There were just CPUs, period. As time progressed, chip designers naturally asked themselves how could they increase performance? And what did they find?

For desktop CPUs, performance was best increased by increasing clock rates.
For server CPUs, performance was best increased by increased core counts.

This is why every server CPU today has 4, 8, or or even 16X as many cores as a desktop CPU -- but much lower clock speeds. I've proven this repeatedly with real-world examples.

Honestly, why do you keep arguing against such basic, self-evident real world facts? Every chip designer in the world disagrees with you. Please: stop the nonsense:


Tell us again your absurd fantasy that the NYSE is secretly building their own custom CPUs in hidden underground laboratories.

Now, to once again correct you:

"Underneath the impressive trading floors of the NYSE, a powerful system of servers and processors hums away — processing millions of quotes and trades every second ...Instead of expensive, high-end machines, the NYSE turned to HP ProLiant servers — over 200 four-way HP ProLiant DL585s and 400 ProLiant BL 685c blade servers, all running RHEL...."

How NYSE moved to Linux and Reduced infra cod.

In the fast-paced world of global finance, the New York Stock Exchange (NYSE) stands as an iconic symbol of capitalism and financial power…

overflowbyte.medium.com

"The NYSE and its subsidiaries conduct trading in the United States and five European countries..... Instead of running on proprietary hardware, the exchange runs on 200 four-way HP ProLiant DL585 servers and 400 ProLiant BL 685c blade servers...."

New York Stock Exchange Runs Trades On Red Hat Linux | InformationWeek

The conversion to Linux followed the acquisition of the Euronext exchange in 2007, and the open source operating system is now powering the NYSE's mission-critical trading systems.

www.informationweek.com

Want to guess what CPUs HP ProLiant servers support? (hint: they're not made in underground Arctic laboratories)

"In the early days of computing, there were no "desktop" or "server" CPUs." - the most wtf argument I've heard. ok grandpa, sit or you'll have a heart attack

"For desktop CPUs, performance was best increased by increasing clock rates." so we don't have tens of cores on desktop? and we are also not seeing a big increase in cores next gen? ok grandpa.

"self-evident real world facts" - yeah, the facts you ignore or simply pretend not to know. I gave exact examples with undeniable proof. All you know to do is lie.

As for the "servers" you mention, you do realize that you are quoting something from 2 decades ago, right?
May 14, 2008

New York Stock Exchange Runs Trades On Red Hat Linux | InformationWeek

The conversion to Linux followed the acquisition of the Euronext exchange in 2007, and the open source operating system is now powering the NYSE's mission-critical trading systems.

www.informationweek.com

The freaking video you linked is from 2008:

And as a history lesson for you grandpa: the NYSE DL585 G5 had Opteron CPUs from the 8300 series starting with the high clocked 95W 2.2GHz Opteron 8354, one the best and most popular high performance server AMD CPUs you could buy. Only a handful Barcelona Opteron CPUs were clocked slightly higher, which were either not available at the time they bought the systems or running too hot and loud and they required the best stability for the servers (like the 119W Opteron 8360 SE (B3) which large data centers avoided because of the heat and how loud the fans were, with too little performance gains).

This is just pathetic... seriously, this is just pathetic. You can literally google or even use your favorite AI to fact check what I wrote. And here you are using 2008 articles that you didn't try to check wtf they are talking about.

 

[Endymio]

"In the early days of computing, there were no "desktop" or "server" CPUs." - the most wtf argument I've heard. ok grandpa, sit or you'll have a heart attack

You realize facts don't change depending on the age of the person expressing them, right? I'm certainly more proud to be a grandfather than a Romanian pig herder.

"For desktop CPUs, performance was best increased by increasing clock rates." so we don't have tens of cores on desktop?

Was that a joke? Your average desktop CPU has 8-24 cores. Server CPUs go up to 288 cores.

"self-evident real world facts" - yeah, the facts you ignore or simply pretend not to know. I gave exact examples with undeniable proof.

LOL, where? I know pig herders drink a lot of vodka at night, but you haven't presented any facts to counter the claim. Server CPUs have more cores and lower clock rates -- because for server workloads, cores > clocks. Why keep denying something so elementary?

As for the "servers" you mention, you do realize that you are quoting something from 2 decades ago, right?
May 14, 2008

You really can't read plain English, can you? My first link was from 2008, when they first moved to those ProLiant servers. The other link dates from 2024. Here it is again:

How NYSE moved to Linux and Reduced infra cod.

In the fast-paced world of global finance, the New York Stock Exchange (NYSE) stands as an iconic symbol of capitalism and financial power…

overflowbyte.medium.com

But please, tell us again how you think the NYSE is secretly building their own CPU chips in underground labs. Of all the absurd statements you've made, that one has to be the most hil

 

[MiqWalsh]

Oh, for Chrissakes. Go and get a room, you two......and give the rest of us some peace & quiet, will ya? Jeezus H...

You're BOTH giving visitors to the site completely the wrong impression.


Miq.