Discussion Should MLID be considered a legitimate source

Mopetar · Feb 14, 2024

A legitimate source of what?

Hot air. Certainly.

Tech rumors. No, not really.

I don't think there's anything wrong with posting some rumor he's reporting on, but expecting it to have any truth behind isn't smart. Healing crystals have a better track record than he does.

NTMBK · Feb 14, 2024

coercitiv said:
In the course of ~6 months his info about the bigger BMG die has gone from 362mm2 on TSMC 4nm to 387mm2 on TSMC 5nm. Giving this type of exact information and then changing with another exact number should be a red flag for anyone with an engineering or financial education. Someone who's merely speculating on the die size would come up with something like 350-400mm2 or ~370mm2.

I mean, sometimes completed dies do get canned, either for economic or performance reasons. I personally sat in a presentation from Nvidia reps about an upcoming "big Maxwell" that never came to market- it had much higher FP64 throughput, and from the power consumption and FLOPS numbers it must have been a 20nm part, not 28nm, even though the slide never mentioned a specific process node. But that chip never actually came to market- we had to wait until Pascal for that level of performance.

Timorous · Feb 14, 2024

Surely all leaks should just be considered speculation bait. Is it really different to transfer silly season in sports where you have rumours flying around about huge moves and then on the odd occasion you actually get something legit like Lewis to Ferrari.

I just use it as a jumping off point to talk nonsense untill more solid info comes out.

coercitiv · Feb 14, 2024

Timorous said:
Surely all leaks should just be considered speculation bait.

Some leakers have much better track records than others.

Years ago somebody went trough the trouble of attempting to check track records for known leakers at the time. Obviously this was a mammoth task, prone to errors of all kinds, but informative nonetheless. Notice how some leakers got over 75%+ accuracy on their claims, while other struggle around 50%. There's people there with 9/10 accuracy. I would argue they are clearly outside of "speculation bait" territory.

dullard · Feb 14, 2024

coercitiv said:
Some leakers have much better track records than others.

Years ago somebody went trough the trouble of attempting to check track records for known leakers at the time. Obviously this was a mammoth task, prone to errors of all kinds, but informative nonetheless. Notice how some leakers got over 75%+ accuracy on their claims, while other struggle around 50%. There's people there with 9/10 accuracy. I would argue they are clearly outside of "speculation bait" territory.

Thanks! That table is exactly what I hoped this thread would become, since I had not seen that table before. I just wanted a way for us to debate accuracy of posts/links/images/videos various people have. Not to debate who leaked what first or personalities/backtracking/blah blah blah.

Timorous · Feb 14, 2024

coercitiv said:
Some leakers have much better track records than others.

Years ago somebody went trough the trouble of attempting to check track records for known leakers at the time. Obviously this was a mammoth task, prone to errors of all kinds, but informative nonetheless. Notice how some leakers got over 75%+ accuracy on their claims, while other struggle around 50%. There's people there with 9/10 accuracy. I would argue they are clearly outside of "speculation bait" territory.

View attachment 93862

That kind of analysis really needs an indication of how far before launch or announcement a claim was made.

Some people only leak a lot closer to launch so their track record is bound to be better.

moinmoin · Feb 14, 2024

igor_kavinski said:
You can learn a lot of things watching him talk. Like how to BS your way out of anything. Very useful skill in the real world.

You mean MLID rivals AI at their core competency? How many FLOPS does he offer?

igor_kavinski · Feb 14, 2024

FlameTail said:
But don't Intel/AMD iGPUs do it already?

Intel yes but not sure about AMD.

Saylick · Feb 14, 2024

moinmoin said:
You mean MLID rivals AI at their core competency? How many FLOPS does he offer?

Depending on who you ask, MLID flops quite a lot.

Hulk · Feb 14, 2024

Remember when the WWF "World Wrestling Federation" changed it's name to WWE "World Wrestling Entertainment?" They basically gave up the game and said, "Yeah it's scripted, but it's still fun!"

MLID should do the same and just say, "Yeah my "inside" information isn't so inside, but I'm still fun!"

FlameTail · Feb 14, 2024

igor_kavinski said:
Intel yes but not sure about AMD.

Why is AMD so behind the curve?

Has Qualcomm already implemented this in 8cx Gen 3?

igor_kavinski · Feb 19, 2024

I think M$ did this with AMD for the original Xbox. At the last minute, they went with Intel Celeron and a Geforce 3 type of GPU.

Hardware Behind the Consoles - Part I: Microsoft's Xbox

www.anandtech.com

The CPU that powers the Xbox is a Coppermine based Pentium III with only 128KB L2 cache. While this would make many think that the processor is indeed a Celeron, one of the key performance factors of the Pentium III that is lost in the Celeron core was left intact for this core. The Coppermine core was left with an 8-way set associative L2 cache instead of the 4-way set associative cache of the Celeron. Based on what we've seen with the Coppermine and Coppermine128 (Celeron) cores we estimate that the 8-way set associative L2 cache gives this particular core a 10% performance advantage over the Coppermine128 core of the Celeron.

So it was a castrated P3 but 10% faster than a Celeron.

Regarding the "leak", true or not, it's good entertainment!

Tigerick · Mar 11, 2024

uzzi38 said:
There's a major feature of STX-Halo that hasn't leaked yet so I'd assume no-one has got actual slides on the part yet. At least not detailed enough slides that I'd be confident in all of the core parts of the thing (don't ask me what it is, when you find out, you'll know what I'm referring to).

Just setting expectations here.

Could STX Halo be supporting LPDDR6 with RDNA5 when launch next year?

FlameTail · Mar 11, 2024

Tigerick said:
Could STX Halo be supporting LPDDR6 with RDNA5 when launch next year?

Unlikely.

Probably LPDDR5X-8533 or LPDDR5X/5T-9600 at best.

uzzi38 · Mar 11, 2024

Tigerick said:
Could STX Halo be supporting LPDDR6 with RDNA5 when launch next year?

Hell no, what made you think THAT would be it?!?!

Tigerick · Mar 11, 2024

uzzi38 said:
Hell no, what made you think THAT would be it?!?!

Cause of delay and release timings of LPDDR6. Anyway, thanks for let me know..

Heartbreaker · Mar 12, 2024

IMO, MLID is for people that want their tech rumors in the form of an embellished soap opera, full of fake drama for clicks.

I could never stand listening to the guy. He just set off my BS detector immediately.

I have no doubt he sometimes gets leaked info sent to him, but he mixes in a bunch of blowhard speculation and fake drama. Plus why would I want tidbits of text based rumor info in video form, so I literally blocked his channel in YT.

I do follow Videocardz which tends to report two things. Selected rumors sourced from other people (including MLID) tagged as rumors, and also stuff they confirmed themselves. When VC says they confirmed something, it's almost always true.

mderbarimdiger · Mar 12, 2024

Ok, looks like we can find common ground after all! @Hitman928 scored 20 likes for breaking down why we shouldn't take MLID as a legit source.

how many other posts do we have racking up 20 likes?

mderbarimdiger · Mar 12, 2024

Timorous said:
That kind of analysis really needs an indication of how far before launch or announcement a claim was made.

Some people only leak a lot closer to launch so their track record is bound to be better.

Yeah.. few people only make claims when they are certain its accurate, so their track record is bound to be better.

Other leakers simply dont care about their track record.

DavidC1 · Mar 12, 2024

MLID is like ChatGPT -

-He can be accurate but is often wrong
-Despite some being wrong, he says with complete confidence
-Just like ChatGPT his "training data" comes from others, not his own
-Sometimes outright hallucinates

Unlike ChatGPT, rather than admitting wrong, he deletes his videos that were wrong.

gdansk · Mar 12, 2024

ChatGPT was initially trained on "real" data.
MLID? Not so much. MLID has intent and reason to make things up (views). ChatGPT does it only because the words are associated together.

Worse than ChatGPT in my judgment.

igor_kavinski · Mar 12, 2024

In other words, he's a waste of precious resources!

branch_suggestion · Apr 5, 2024

Is Tom okay?

adroc_thurston · Apr 5, 2024

branch_suggestion said:
Is Tom okay?

put JUST hair on him.
It's never been more over for him, got catfished in the open.

branch_suggestion · Apr 5, 2024

adroc_thurston said:
View attachment 96397
put JUST hair on him.
It's never been more over for him, got catfished in the open.

I'm sure people have the evidence saved before he inevitably nukes it, he lost the game.

Discussion Should MLID be considered a legitimate source

Diamond Member

Lifer

Golden Member

Diamond Member

Elite Member

Golden Member

Diamond Member

Lifer

Diamond Member

Diamond Member

Platinum Member

Lifer

Senior member

Platinum Member

Platinum Member

Senior member

Diamond Member

Member

Member

Member

Platinum Member

Lifer

Member

Platinum Member

Member