AMD Threadripper Delidded, with a Multi-Core Surprise Under the Hood
extremetech.com
extremetech.com
I can understand them not wanting to let these new lines/prices be the norm too fast and lose money on margins, but I certainly did not expect them to be caught in what can only be total surprise and unpreparedness.
If there was a monopoly over the last decade it was only because AMD wasn't producing a competitive offering. It was not because of marketplace abuses like Intel's been accused of in the past.
source?
In my experiance that is false.
the Space heaters what where the P4's where unstable and had continual problems compared to the Athons. I almost never had a problem out of a AMD based system but Intel based was a constant stability battle
The Core line up changed that but it was not until after the Core2Duo's that Intel really started to surpass AMD
>If there was a monopoly over the last decade it was only because AMD wasn't producing a competitive offering. It was not because of marketplace abuses like Intel's been accused of in the past.
This is also objectively false for any market outside the DIY/Gamer space. Intel made known, and proven anti-competitive deals with many OEM PC builders to ensure they would not offer AMD Chips to their customers
Ref: 2010 FTC Anti-Trust Settlement
Then you have the whole Patent battle of x86 itself which created the Douoply of AMD/Intel excluding all others from the market which Intel would have loved to have to itself asserting patent rights on the various x86 technologies.
So no you can not say that Intel is on top only because they have/had an objectively better product, though out Intel entire existence they have attempted to everything in their power (and some of it proved to not be so legal) to kill any competition to their market share not with innovation but with contracts, patents and other uses of the legal system
They continue this pattern today with their not so veiled threat against Qualcomm and MS over having x86 apps run on ARM while at the same time attempting to Sue Qualcomm for their aggressive enforcement of patents and restricted licensing of same
The "surprise" is that there are four dies in the CPU, two of which appear to be unused (I'm not sure how they reached that conclusion, I can't see the video).
It's not particularly surprising or novel though, and there can be many reasonable explanation about why they would do such a thing and the article doesn't reach a conclusion. Silicon is cheap, maybe those are non-functional dies just here for mechanical reasons.
Have seen the video. Der8auer planned a test to ascertain this, but was unable. AMD later confirmed it to him.
Then you consider this is their answer to AMD's new chips, cutting corners...
They've been overclocking hostile for almost two decades now (to varying degrees at varying times), seemingly make no efforts towards cross-generation socket compatibility, and are now using PCIe I/O connectivity as a club to drive buyers to higher-margin chips.
Aside from the K chips and the unlocked Pentiums, I'm hard-pressed to think of anything Intel has intentionally done to cater to the enthusiast market in recent memory. And even K-chips are a token gesture in response to a problem they created.
It looks to me like the entirety of Zen from $100 to $1000 CPUs is the same design, the same manufacturing process, the same chips, just bucketed according to speed test and failed cores and arranged into MCM packaging.
Buy early if you want to profit from this, if that turns out to be possible.
When compiling a large J2EE project in NetBeans, it is slightly faster than my FX 6100, with both systems running solid-state drives.
However. Soldering these is quite difficult with smaller dice and thermal cycling can result in cracks which impacts reliability, because silicon, the solder and the IHS have different thermal expansion coefficients. There are multiple processing steps required and the solder used is of course special as well. (In fact, in the article you can see the gold plated areas on the nickel-plated IHS where the dice make contact).
TIM (i.e. thermal paste) is very easy to apply and has a predictable lifespan pretty much regardless of die size. The thermal resistance is much worse, though.
The soldering process requires precise thermal control and cycle control. Get the solder just a little too hot or put a little too much on the die and you'll overheat it. Heat or cool it down too fast or far and it'll crack. Compare that with what is needed to squirt a pea of thermal paste on the die and slap the lid on and you'll get why it makes sense. Die yields are already low enough with the latest ultra-tiny 14nm-and-below processes, losing a good die to the "easy" part of CPU manufacturing (assembly) is very costly.
I mention this because all these AMD threads are making me get the itch. And it might be a good idea to support some competition with Intel.
Can anyone point me to a good information resource for these new chips?
For threadripper I would wait a bit till the release articles roll out. That release will be covered extensively. But what is expected is "just" a bigger Ryzen processor with more cores and maybe some latency issues, tbd.
http://www.gamersnexus.net/hwreviews/2822-amd-ryzen-r7-1800x...
http://www.gamersnexus.net/hwreviews/2875-amd-r5-1600x-1500x...
http://www.gamersnexus.net/hwreviews/3002-amd-r3-1200-review...
Seems AMD retracted permission: https://youtu.be/JbffGyrkjMM
Managed to see it before it was taken away. I wonder why AMD changed their mind on this... Could Threadripper and EPYC be cross-compatible after all?
What I was suggesting was that new hardware always comes out and it's always bigger/faster (and even 2x improvement isn't even historically that much). Are the people upvoting this because they think every single new piece of hardware is newsworthy, or that there's something special here (e.g. they feel intel has a monopoly and are invested in AMDs success), or are they shill accounts ?
> or are they shill accounts
If you know something the mods don't, email them. But that's not a particularly nice thing to insinuate.
Additionally:
> Please don't submit comments complaining that a submission is inappropriate for the site.
Whether your original comment was meant to indicate complaint, it certainly comes across that way.
Site guidelines: https://news.ycombinator.com/newsguidelines.html
However that same logic couldn't really justify the downvoting of my follow-up "could anybody who downvoted this please explain why?"