Intel Xeon 6980P "Granite Rapids" Linux Benchmarks

[sophisticles]

I don't care about the level of your intelligence, but the fact you call Threadripper a scam (and other equally nonsensical statements of yours) says a lot about it LOL

Allow me to reiterate, Threadrippers are a scam and evidently there are enough pigeons out there that AMD has been able to see sustained growth the past number of years.

But all scams, no matter how good or believable, eventually come to an end, and TR is no different.

As for the mental health, I was referring to your post in this forum where your openly declared yourself that you have some kind of issues regarding it. I won't waste my time searching for it now though, but I'm not lying.

I have never said that I have mental health issues.

What I have said is that I used to work for two hospitals decades ago, one of them was in behavioral health, namely PICU, ACDS and ACDU.

People that lack the knowledge and education to offer cogent counter arguments to my stance on the GPL, S76 and similar issues have attempted to disparage my opinions by misrepresenting my background.

The remaining of your post is a diarrhea of buzzwords looking like a chatGPT answer. Show the source of your software or just leave this topic.

You know how I feel about working for free.

I have no intention of providing the source or a binary; why would I?

In all fairness, if someone has a programming background, and a data analysis background and a math background and knows what they are doing, should be able to replicate the functionality.

But here's the best part, I have made a public prediction and I am willing to further define my prediction, I say that Intel goes to 27 a share by the end of October this year.

Bookmark this thread and wait until Oct 31 2024 to see if Intel has reached 27.

If yes, then maybe I know what I am talking about.

If no, then it's diarrhea of buzzwords.

[JediMindTricks]

His comment was actually useful especially to newbies at investing. As in, he's mostly right in general way.

In general you buy stocks when they're low and hit hard by people panicking to sell, not stocks that are sky high. Buy Low, Sell High.

Or Buy Low and hold. But you still want to buy low so you get more shares (and thus dividends). Simple logic.

While I agree with this idea in general. INTC has never been a great stock. The SP500 index has historically outperformed it. I'd encourage newbies to stay in index funds. But if someone really wants to buy INTC make sure you buy below $25 and don't buy all shares at once, dollar cost average.

[trifud]

Impressive performance. Hope it stays the same after a few rounds of vulnerability fixing.

[timo]

It's been said that intelligence is rare and that assessment is correct, as evidenced by your comment.

The software is custom software that I wrote that uses traditional analysis techniques specific to stocks, coupled with analysis technologies used for data analysis and AI powered analysis.

I used 50 and 200 day SMA, 12 and 26 day EMA, Bollinger Bands, Ichimoku Cloud, as well as some custom SMA and EMA.

Additionally I used Linear, Polynomial, Random Forest, Decision Tree and a couple of other regressions to analyze historical closing prices and build a model.

Finally I added Scikit-Learn, Statsmodel, Prophet, NueralProphet, IBM's Watson AI library, PyTorch, and TensorFlow analysis.

Using each method I trained various models on the historical data, using time frames of various sizes, and then used the resulting models to make predictions for a specific time frame into the future,

I then compared the predictions against the actual data and calculated a percent accuracy.

All models, if you vary the range, can achieve a 95% accuracy, if you narrow the range, some models, on some stocks, can actually achieve a 98.5% accuracy, meaning the predicted price for a given day is within 98.5% of what actually happens.

But by all means, do not listen to me.

I would find it very enjoyable knowing that I gave you quality information, but due to idiocy on your part you ignored it and then knowing that you were kicking yourself when other people laughed all the way to the bank.



Who with a medical degree declared this?

What bothers you is that I am way smarter than you and you, to your credit, do have enough brain power to realize it, and it bothers you.

It's OK, that's the way life is.

You can use all the fancy methods and techniques you want but it's just gambling, be it informed or not. Everyone in the business knows this, and there is no magic, full stop.
For one that wins there are hundreds that lose (typically those that have no access to privileged information). The business is rot to the bone.

By the way, from the way you define prediction accuracy you have shown us that you understand little of time series analysis and of the methods you mention. Let me guess: you do not have a proper background in statistics.

[baka0815]

Michael
"Password123!" is not a secure password.

[drakonas777]

Allow me to reiterate, Threadrippers are a scam and evidently there are enough pigeons out there that AMD has been able to see sustained growth the past number of years.

Here is a definition of scam: https://dictionary.cambridge.org/dic...y/english/scam

Threadripper is objectively and logically not a scam. You are either using this word incorrectly or using hyperbolic expression to attack AMD for some emotional reason. A sign of weak mind in both cases

Anyway, I'm done with this offtopic.

[Paradigm Shifter]

I was pretty put off when CPUs hit 300W TDP... 500W is a whole different beast; what are the cooling solutions like for this? I imagine there is next to no chance of seeing these in 2U servers.

[coder]

I imagine there is next to no chance of seeing these in 2U servers.

With water cooling, there is. There've been water-cooled 2U HPC machines that burn a lot more power per box than these do.

[Anux]

"Password123!" is not a secure password.

Exactly, I use "Pissw0rd321?" you should too. That is much more secure!
Even better let's have a poll where everyone posts their passwords and then we select the most secure one and everyone uses that from now on.

Mmmh, about power efficiency, I don't know... the 2P 6980P processors have a 1kw TDP summed up

TDP is not energy consumption, but as you said we will see when the tests are there. I was actually surprised that they even surpassed Zen 4 performance so I don't rule out that they have massive improvements in power consumption too.

Interestingly enough, after disappearing from the upcoming desktop core 200 series, hyperthreading is back here Something that should be asked to all those Intel-loving people that was continously repeating like a mantra SMT was bad

It was never a question for server, because heavy multi threaded workloads benefit the most from HT.

It's being compared against their previous generation. Intel on a 3nm-class node vs. AMD on a 5nm-class node. It's no wonder that it beats EPYCs with fewer or smaller cores.

A smaller node is not a guaranty but I would also bet that Zen 5 might be roughly the same performance but slightly ahead in efficiency.

Zen 5 doesn't increase cache per-core, so Zen 5 probably won't do enough to overturn those results.

I heard that the X3D cache size might increase, we will see.

I was pretty put off when CPUs hit 300W TDP... 500W is a whole different beast; what are the cooling solutions like for this? I imagine there is next to no chance of seeing these in 2U servers.

There are already water cooled 2U systems out there as well as whole water cooled HPC data centers. Last gen already had 2 x 350W in air cooled 2U. In the server space it is all about density and efficiency, noise of the cooling system is the last thing to worry about.

[Michael]

Michael
"Password123!" is not a secure password.

Those were Intel's passwords.