Chinese semiconductor industry

Status
Not open for further replies.
tphuang

tphuang

Lieutenant General
Staff member
Super Moderator
VIP Professional
Registered Member
Performance of Loongson 2K2000 against its competition. Keep in mind, 2K2000 runs on 28 nm process vs 2K1000LA at 40 nm and Phytium E2000Q at 14nm. So, E2000Q is able to run at higher frequency than 2K2000 and use more advanced process but has worse performance. 2K2000 is now able to have 3x the performance of 2K1000LA despite keeping same number of cores and just using a more advanced process. This indicates the updated LoongArch ISA and LA364 core is quite competitive vs Phytium's ARM design. for lower power consumption applications.

Please, Log in or Register to view URLs content!

In terms of power consumption
Please, Log in or Register to view URLs content!
E2000Q is at 6.4W
Please, Log in or Register to view URLs content!
2K2000 is at 4W

Keep in mind that Loongson has also completed development of LA664 core for its next gen 3A6000 CPU which is supposed to be comparable in performance to AMD Zen 3 despite using 12nm process vs 7nm for Zen 3.
Please, Log in or Register to view URLs content!
I think there is a lot of value in continued development of LoongArch ISA and Loongson Core.
 
O

Orthan

Senior Member
Seems that this hasnt been posted here yet. Reuters article about chip curbs against china. Its almost all behind a paywall, but it mentions that there are no immediate chip curbs against china by the netherlands and japan.

Please, Log in or Register to view URLs content!


For those that have acess to the whole article, what does it say?
 
daifo

daifo

Captain
Registered Member
Orthan said:
Seems that this hasnt been posted here yet. Reuters article about chip curbs against china. Its almost all behind a paywall, but it mentions that there are no immediate chip curbs against china by the netherlands and japan.

Please, Log in or Register to view URLs content!


For those that have acess to the whole article, what does it say?
Click to expand...

Basically what you wrote, Japan and the Dutch are not convince (yet)
Same article via yahoo without the paywall
Please, Log in or Register to view URLs content!
 
FairAndUnbiased

FairAndUnbiased

Brigadier
Registered Member
Orthan said:
Seems that this hasnt been posted here yet. Reuters article about chip curbs against china. Its almost all behind a paywall, but it mentions that there are no immediate chip curbs against china by the netherlands and japan.

Please, Log in or Register to view URLs content!


For those that have acess to the whole article, what does it say?
Click to expand...
It's an irrelevant political propaganda article for non SMEs because it confuses semiconductors with semiconductor fabrication equipment. Japan and Netherlands don't have leading edge semiconductor foundries, so what chip curbs can there be? If they're talking about equipment then why is it a "chip curb" and not a "tool takedown"?
 
O

Overbom

Brigadier
Registered Member
FairAndUnbiased said:
It's an irrelevant political propaganda article for non SMEs because it confuses semiconductors with semiconductor fabrication equipment. Japan and Netherlands don't have leading edge semiconductor foundries, so what chip curbs can there be? If they're talking about equipment then why is it a "chip curb" and not a "tool takedown"?
Click to expand...
You should know by now how western reporting works.

What I get from the article is that the US still hasn't managed to convince/force Netherlands and Japan to follow along with semiconductor (equipment) restrictions. So, January should be an extra month for China.

To reiterate, imo there is no doubt in my mind that the restrictions will happen. What's the issue here is when and how expansive these sanctions will be.

One month of the US not reaching a deal with Netherlands and Japan is one extra month for Chinese fabs to get as much semi equipment as possible from those countries
 
olalavn

olalavn

Senior Member
Registered Member
Orthan said:
Seems that this hasnt been posted here yet. Reuters article about chip curbs against china. Its almost all behind a paywall, but it mentions that there are no immediate chip curbs against china by the netherlands and japan.

Please, Log in or Register to view URLs content!


For those that have acess to the whole article, what does it say?
Click to expand...
Do they mention production equipment? no? so ASML, Japan has the right to sell equipment for them
 
T

tokenanalyst

Brigadier
Registered Member
FairAndUnbiased said:
X ray has much more relevant information than optical since nobody cares about the optical parameters but elemental analysis, in situ density measurement and film roughness are all critically important and would need to be measured elsewhere anyhow.

Most of all, XRF is almost entirely model independent. You see what you get. Optical metrology is highly model dependent, picking wrong parameters, having problems with software or having an unexpected film stack are all big problems. XRR is model dependent too, but is well characterized.

In addition, X-ray metrology can be extended to XRD for strain measurements for strain engineered interfaces with no further specialized environmental conditions. There are already combined XRD, XRR and XRF instruments for total analysis of thin films.
Click to expand...
X-RayF you know if you deposit the right material on the right amount based on preloaded models. With X-Ray reflectivity you get the features of the layer, with X-Ray diffraction is like X-ray crystallography you can analyze the molecular structure of the material. Very interesting that they combined all three features in one tool.
 
FairAndUnbiased

FairAndUnbiased

Brigadier
Registered Member
tokenanalyst said:
X-RayF you know if you deposit the right material on the right amount based on preloaded models. With X-Ray reflectivity you get the features of the layer, with X-Ray diffraction is like X-ray crystallography you can analyze the molecular structure of the material. Very interesting that they combined all three features in one tool.
Click to expand...
You actually don't need too many models for bulk elemental analysis, XRF is used even for analyzing elemental composition of rocks. Only when you want a nondestructive depth profile do you need significant modeling, but that gets really complicated fast.

I think it's not too hard to stack all the equipment in 1 package. You have 1 source but separate detectors for each: a topside detector for XRF at a near 90 degree angle to the incident beam since fluorescence is isotropic but incident beam undergoes low angle specular (GIXRF) or total internal reflection (TXRF geometry) and you can ignore the incident beam, then you have a 1D array detector for XRR and XRD.
 
Status
Not open for further replies.
Top