Semiwiki on LinkedIn Semiwiki on Facebook Semiwiki on Twitter Register / Log in 
Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/threads/amd-says-intels-horrible-product-is-causing-ryzen-9-9800x3d-shortages.21836/page-2
        )

    [addOns] => Array
        (
            [DL6/MLTP] => 13
            [Hampel/TimeZoneDebug] => 1000070
            [SV/ChangePostDate] => 2010200
            [SemiWiki/Newsletter] => 1000010
            [SemiWiki/WPMenu] => 1000010
            [SemiWiki/XPressExtend] => 1000010
            [ThemeHouse/XLink] => 1000970
            [ThemeHouse/XPress] => 1010570
            [XF] => 2021770
            [XFI] => 1050270
        )

    [wordpress] => /var/www/html
)

AMD says Intel's 'horrible product' is causing Ryzen 9 9800X3D shortages

"We knew we built a great part. We didn't know the competitor (Intel) had built a horrible one," quipped AMD executive Frank Azor. "So the demand has been a little higher than we forecasted."

This is really silly. Looks like AMD is also blaming their own non X3D parts if we go by this statement. Are their 9950X & 9900X competitive with their 9800x3D gaming CPU?

The fact of the matter is Intel does not have an answer to AMD's x3D chips in gaming. Even if Intel improves the ARL-S latency issues iwth, they still would not be able to beat 9800X3D in gaming so does AMD's own Zen 5 16 core processors (like 9950X)!!! Refer to the following review of ARL-S in content creation and productivity work. They are competitive with Zen 5 (win some, lose some). Intel has done a good job on improving perf/watt and reducing temperature (both of these things were complaints about Intel processors for some time)

www.pugetsystems.com

Intel Core Ultra 200S Content Creation Review

We examine the performance of Intel's new Core Ultra 200S series desktop processors in content creation applications.
www.pugetsystems.com www.pugetsystems.com
www.pugetsystems.com

Power Draw, Cooling, and Efficiency: Intel Core Ultra 200S

In this article, we examine the power draw and efficiency of the new Intel Core Ultra 200S series processors.
www.pugetsystems.com www.pugetsystems.com

1736544885029.png


Now lets looks at 9800X3D in productivity workloads (nobody should buy them for this workloads imo but just for the sake of comparison here)
www.pugetsystems.com

AMD Ryzen 7 9800X3D Content Creation Review

We examine the performance of AMD's new Ryzen 7 9800X3D in content creation applications.
www.pugetsystems.com www.pugetsystems.com
1736544975118.png


Does this make 9800x3D a horrible product!!!!?

Intel needs a response to AMD's Vcache to be competitive in gaming. Online rumor is its coming in next generation for Nova Lake. Only thing is, will Intel survive the mindshare loss among gamers because of the current situation!
 
Intel demonstrated running XeSS 2 on Arrow Lake H using the NPU in a very low power and quiet mode:

"As expected, these AI-centric processors also have an in-built NPU that delivers up to 11 TOPS"
www.tomshardware.com

Intel launches Arrow Lake mobile family with Core Ultra 200HX and 200H processors for gaming laptops

Shooting Arrows into the mobile market.
www.tomshardware.com www.tomshardware.com

The NPU TOPS is only 11! I assume it is INT8. I did a video yesterday on how to read TOPS...

www.theverge.com

Microsoft explains how its DLSS competitor uses AI to improve any game

Microsoft explains how Auto SR differs to DLSS.
www.theverge.com www.theverge.com
 
siliconbruh999 said:
Intel ARL supply seems to be low from the stock I guess it makes sense they don't want to give TSMC Revenue but they don't have options
Click to expand...
No, this doesn't make sense. Intel has already invested hundreds of millions in developing Arrow Lake, and some production costs, like photo masks, are fixed. To lower the average cost per die, large volumes are necessary. Additionally, if the products are competitive, Intel should earn more profit from selling the entire CPU than TSMC does from making the dies.

Intel will either place substantial orders with TSMC expecting strong sales or not place orders at all. There's no way Intel would design for TSMC's N3B process and hope for low volumes. Intel isn't that wealthy.
 
Maximus said:
No, this doesn't make sense. Intel has already invested hundreds of millions in developing Arrow Lake, and some production costs, like photo masks, are fixed. To lower the average cost per die, large volumes are necessary. Additionally, if the products are competitive, Intel should earn more profit from selling the entire CPU than TSMC does from making the dies.
Click to expand...
But we are seeing such a low volume of ARL including ARL-S when a Intel product is launched in the past 3 years there are ample amount to be bought it's not the case with ARL-S some people are not seeing in stock and if you are saying it's hot selling that's not true as well
Maximus said:
Intel will either place substantial orders with TSMC expecting strong sales or not place orders at all. There's no way Intel would design for TSMC's N3B process and hope for low volumes. Intel isn't that wealthy.
Click to expand...
Or they are doing some shenanigans idk but this time volume is low there were not that many ARL SKU at CES as well there were 5 Panther Lake reference design!

Is it possible to change the flavor of node you want ?
For example I bought N3B but I would like to change some of that N3B capacity to N3E is it possible through negotiations?
 
Last edited:
siliconbruh999 said:
But we are seeing such a low volume of ARL including ARL-S when a Intel product is launched in the past 3 years there are ample amount to be bought it's not the case with ARL-S some people are not seeing in stock and if you are saying it's hot selling that's not true as well
Click to expand...
I'm not suggesting that Arrow Lake is a hot seller. What I'm saying is that Intel must have made a strong business case when planning Arrow Lake. If not, it indicates a bigger issue with Intel's product planning than with their foundry. (Though feedback from Lunar Lake indeed suggests significant problems with Intel's product planning.)

This implies that the low sales of Arrow Lake were unexpected for Intel, at least during the initial stages of its development.
siliconbruh999 said:
Is it possible to change the flavor of node you want ?
For example I bought N3B but I would like to change some of that N3B capacity to N3E is it possible through negotiations?
Click to expand...
It depends on how much extra money you're willing to spend and the utilization rate of process nodes. Currently, TSMC's N3 and N5 nodes are operating at nearly 100% capacity, making it difficult to change orders for these nodes. Additionally, every change incurs a cost. TSMC has been in the foundry business for over three decades and has developed comprehensive mechanisms to prevent customers from frequently changing their orders, unlike a Costco return policy.

The bigger issue lies in the design. N3B and N3E are not design-compatible. Transforming a design from N3B to N3E would require enormous time and effort.
 
dkr1986 said:
"We knew we built a great part. We didn't know the competitor (Intel) had built a horrible one," quipped AMD executive Frank Azor. "So the demand has been a little higher than we forecasted."

This is really silly. Looks like AMD is also blaming their own non X3D parts if we go by this statement. Are their 9950X & 9900X competitive with their 9800x3D gaming CPU?

The fact of the matter is Intel does not have an answer to AMD's x3D chips in gaming. Even if Intel improves the ARL-S latency issues iwth, they still would not be able to beat 9800X3D in gaming so does AMD's own Zen 5 16 core processors (like 9950X)!!! Refer to the following review of ARL-S in content creation and productivity work. They are competitive with Zen 5 (win some, lose some). Intel has done a good job on improving perf/watt and reducing temperature (both of these things were complaints about Intel processors for some time)

www.pugetsystems.com

Intel Core Ultra 200S Content Creation Review

We examine the performance of Intel's new Core Ultra 200S series desktop processors in content creation applications.
www.pugetsystems.com www.pugetsystems.com
www.pugetsystems.com

Power Draw, Cooling, and Efficiency: Intel Core Ultra 200S

In this article, we examine the power draw and efficiency of the new Intel Core Ultra 200S series processors.
www.pugetsystems.com www.pugetsystems.com

View attachment 2667

Now lets looks at 9800X3D in productivity workloads (nobody should buy them for this workloads imo but just for the sake of comparison here)
www.pugetsystems.com

AMD Ryzen 7 9800X3D Content Creation Review

We examine the performance of AMD's new Ryzen 7 9800X3D in content creation applications.
www.pugetsystems.com www.pugetsystems.com
View attachment 2668

Does this make 9800x3D a horrible product!!!!?

Intel needs a response to AMD's Vcache to be competitive in gaming. Online rumor is its coming in next generation for Nova Lake. Only thing is, will Intel survive the mindshare loss among gamers because of the current situation!
Click to expand...


AMD chips now comprise 55 percent of Puget Systems orders — AMD makes big inroads in professional systems

"In terms of overall CPUs sold, AMD has finally dethroned Intel after four long years, capturing almost 55% of all orders in Q4 2024. Since this number encapsulates both consumer and workstation-grade offerings, let's go over them individually for more insight. Client CPU sales (Ryzen/Core) see...
semiwiki.com
 
siliconbruh999 said:
But we are seeing such a low volume of ARL including ARL-S when a Intel product is launched in the past 3 years there are ample amount to be bought it's not the case with ARL-S some people are not seeing in stock and if you are saying it's hot selling that's not true as well

Or they are doing some shenanigans idk but this time volume is low there were not that many ARL SKU at CES as well there were 5 Panther Lake reference design!

Is it possible to change the flavor of node you want ?
For example I bought N3B but I would like to change some of that N3B capacity to N3E is it possible through negotiations?
Click to expand...
Maximus said:
I'm not suggesting that Arrow Lake is a hot seller. What I'm saying is that Intel must have made a strong business case when planning Arrow Lake. If not, it indicates a bigger issue with Intel's product planning than with their foundry. (Though feedback from Lunar Lake indeed suggests significant problems with Intel's product planning.)

This implies that the low sales of Arrow Lake were unexpected for Intel, at least during the initial stages of its development.

It depends on how much extra money you're willing to spend and the utilization rate of process nodes. Currently, TSMC's N3 and N5 nodes are operating at nearly 100% capacity, making it difficult to change orders for these nodes. Additionally, every change incurs a cost. TSMC has been in the foundry business for over three decades and has developed comprehensive mechanisms to prevent customers from frequently changing their orders, unlike a Costco return policy.

The bigger issue lies in the design. N3B and N3E are not design-compatible. Transforming a design from N3B to N3E would require enormous time and effort.
Click to expand...

Those most important tiles of the Arrow Lake and Lunar Lake are made by TSMC and packaged by Intel. With the relatively low sales volume of Lunar Lake and low availability of Arrow Lake, I started thinking of if there are some issues or bottlenecks in the Intel packaging operations.
 
hist78 said:
Those most important tiles of the Arrow Lake and Lunar Lake are made by TSMC and packaged by Intel. With the relatively low sales volume of Lunar Lake and low availability of Arrow Lake, I started thinking of if there are some issues or bottlenecks in the Intel packaging operations.
Click to expand...
That's what I heard as well: Foveros packaging isn't performing as well as expected.
 
Maximus said:
I'm not suggesting that Arrow Lake is a hot seller. What I'm saying is that Intel must have made a strong business case when planning Arrow Lake. If not, it indicates a bigger issue with Intel's product planning than with their foundry. (Though feedback from Lunar Lake indeed suggests significant problems with Intel's product planning.)

This implies that the low sales of Arrow Lake were unexpected for Intel, at least during the initial stages of its development.

It depends on how much extra money you're willing to spend and the utilization rate of process nodes. Currently, TSMC's N3 and N5 nodes are operating at nearly 100% capacity, making it difficult to change orders for these nodes. Additionally, every change incurs a cost. TSMC has been in the foundry business for over three decades and has developed comprehensive mechanisms to prevent customers from frequently changing their orders, unlike a Costco return policy.

The bigger issue lies in the design. N3B and N3E are not design-compatible. Transforming a design from N3B to N3E would require enormous time and effort.
Click to expand...
I was not talking about design cause I am talking about different designs the iGPU for Panther Lake is already N3E same with Falcon Shores
 
Many comments about the 9800X3D's productivity miss the point. The 9800X3D is designed specifically for 'premium' gaming, so productivity isn't a major concern for its target buyers. Similarly, we wouldn't critique the gaming performance of Epyc or Xeon processors either.

While it's true that the GPU is more important than the CPU for gaming, and pairing a mid-high range CPU with an RTX 4090/5090 can provide about 90% of the experience compared to using the 9800X3D, those who purchase an RTX 4090/5090 are seeking the best gaming CPU since they've already invested significantly more in their GPU than their CPU.

For premium gaming, there's no reason to buy Arrow Lake (Ultra 9 285K) because its gaming performance has regressed compared to the 14900K. From what I recall, Arrow Lake (15th gen) is the first time Intel has experienced a regression in gaming performance compared to the previous generation in recent years.
 
siliconbruh999 said:
I was not talking about design cause I am talking about different designs the iGPU for Panther Lake is already N3E same with Falcon Shores
Click to expand...
My comments remain valid: The negotiation to relocate booking capacity hinges on how much extra money you're willing to spend and the utilization rate of process nodes.

I believe this is precisely why Pat flew to Taiwan multiple times to meet with CC Wei.
 
siliconbruh999 said:
I don't think it's a packing issue Cause MTL is still doing pretty good volume relatively I think it's the margins those bean counters are worried about which ARL just drops
Click to expand...

Unless the net profit margin is negative for a particular product, I don't think Intel would refrain from selling any products right now. It's because one of the serious troubles Intel encountered is its shrinking revenue. Without maintaining or increasing its revenue, Intel's future is very questionable.
 
hist78 said:
Unless the net profit margin is negative for a particular product, I don't think Intel would refrain from selling any products right now. It's because one of the serious troubles Intel encountered is its shrinking revenue. Without maintaining or increasing its revenue, Intel's future is very questionable.
Click to expand...
Definitely they need to increase revenue but it has been decreasing
 
Maximus said:
Initially, AMD viewed Arrow Lake as a formidable competitor due to its superior process node. Additionally, the first wave of the Zen 5 launch did not receive much acclaim, both prompting AMD to expedite the release of the 9800X3D to boost Zen 5 series sales.

Regarding the bottleneck for the 9800X3D, I suspect it is related to packaging. This is because (1) the CCD is shared across the entire Zen 5 series, allowing for easy coordination, (2) the V-cache SRAM is shared by the X3D lineup, also facilitating coordination, (3) TSMC's packaging expansion resources are heavily focused on CoWoS, and (4) the 9800X3D uses second-generation X3D technology, which might require different packaging lines.

If packaging is indeed the bottleneck, the initial quantities of the 9900X3D, 9950X3D, and 9950HX3D could be very limited at launch.
Click to expand...

I think a variant #4 is most likely -- they were still producing the old style X3D solution (packaging, dice, etc.) for Zen 3 and 4 X3D (and related Epyc chips), while ramping up for Zen 5 (and related) style X3D. The variant being the total # of actual "second gen" cache dies produced since they seem to have pulled the schedule forward.
 
Other topic - re: Arrow Lake volume.

We don't know whether Bob Swan intended for N3 to carry Arrow Lake 100%, or if he was still banking on some Arrow Lake capacity internal to Intel (packaging, and "3nm" or whatever the roadmap equivalent to 20A is/was). It's possible that the volume is a little low simply because they didn't order enough capacity.

We also had reports that Gelsinger was going to Taiwan to negotiate TSMC volumes, which could mean anything with respect to Arrow Lake. (More capacity from TSMC, less capacity from TSMC, deferred capacity to another product produced at TSMC, etc..).

Tl;dr too many unknowns on why the volume of Arrow Lake is what it is, other than if it's not ramped up highly in Q2 this year then there's a major problem.
 
You must log in or register to reply here.
Back
Top