• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

AMD Financial Analyst Day 2020 Live Blog

btarunr

Editor & Senior Moderator
Staff member
Joined
Oct 9, 2007
Messages
46,362 (7.68/day)
Location
Hyderabad, India
System Name RBMK-1000
Processor AMD Ryzen 7 5700G
Motherboard ASUS ROG Strix B450-E Gaming
Cooling DeepCool Gammax L240 V2
Memory 2x 8GB G.Skill Sniper X
Video Card(s) Palit GeForce RTX 2080 SUPER GameRock
Storage Western Digital Black NVMe 512GB
Display(s) BenQ 1440p 60 Hz 27-inch
Case Corsair Carbide 100R
Audio Device(s) ASUS SupremeFX S1220A
Power Supply Cooler Master MWE Gold 650W
Mouse ASUS ROG Strix Impact
Keyboard Gamdias Hermes E2
Software Windows 11 Pro
AMD Financial Analyst Day presents an opportunity for AMD to talk straight with the finance industry about the company's current financial health, and a taste of what's to come. Guidance and product teasers made during this time are usually very accurate due to the nature of the audience. In this live blog, we will post information from the Financial Analyst Day 2020 as it unfolds.



20:59 UTC: The event has started as of 1 PM PST. CEO Dr Lisa Su takes stage.





21:05 UTC: AMD details its strategy, that the backbones of its business are "Graphics", "Compute" and "Solutions".


21:07 UTC: High performance computing (including enterprise HPC) is a focus area for AMD.


21:08 UTC: RDNA will last us for the next five years, AMD will continue to invest in their Radeon Graphics RDNA architecture.


21:10 UTC: When AMD entered the market with Ryzen, suddenly performance went up significantly for both segments. Intel's dominance has been holding back progress.


21:11 UTC: AMD highlights how 2019 was a key year for the company. 8% of market share growth per quarter. "Very proud of Department of Energy supercomputer deployment wins."


21:15 UTC: "We're continuing to invest in graphics, winning in 1080p and 1440p". AMD is the exclusive partner for Apple Macs, and lead game consoles.


21:17 UTC: AGR (annual growth rate) detailed. All businesses contributed — roughly 14% year-over-year improvement, every year, since 2015.


21:19 UTC: AMD is very optimistic about the next five years, as they focus on expansion in they key market areas.


21:21 UTC: AMD recognizes the markets its playing with. Datacenter is worth $35 billion each year, and is mostly untapped for AMD. The PC market at $32 billion per year isn't growing much anymore, but it's a huge market that's important for AMD. Gaming (consumer graphics and game consoles): "we this as a good market, lots of people are gaming, these are exciting opportunities for us".


21:22 UTC: Here goes. AMD CEO talks about tech investments on upcoming tech. Zen 3 and RDNA 2 confirmed!


21:23 UTC: AMD talks about Zen 3 on 7 nanometer, and Zen 4 on 5 nm. On the Radeon side, RDNA2 will use 7 nm and RDNA 3 will use an "Advanced Node". Looks like, unlike on the CPU side, they aren't fully decided yet on the fabrication process for RDNA 3. AMD is putting heavy investment in process and packaging, AMD is betting on X3D die-stacking being the next big thing.


21:25 UTC: There is an insatiable demand for compute in the data-center market. Computation model sizes are growing tremendously.


21:26 UTC: For the datacenter AMD is preparing a new architecture called "CDNA", for GPU-based compute accelerators. GPUs without raster hardware is happening.


21:28 UTC: AMD is adding value to the PC market. Note the image in the middle, under "graphics". Is this the new AMD VGA reference design? Looks like the company is finally moving away from blower coolers!


21:30 UTC: AMD has a strategy in place to drive shareholder returns over the next 5 years.


21:32 UTC: What was unrealistic/ambitious in 2017 is realistic now.


21:35 UTC: CTO Mark Papermaster takes stage. Talks about reengineering of the engineering approach at AMD.


21:46 UTC: AMD 7 nm execution detailed.


21:47 UTC: AMD is confident that they process technology will be able to compete with Intel's upcoming 7 nanometer tech.


21:48 UTC: The AMD CPU roadmap is detailed for the years leading up to 2022. Here we see Zen 3 on 7 nanometer, and Zen 4 on 5 nanometer, both at TSMC. AMD is aware that their competition isn't resting, so they're investing a lot into future technologies.


21:49 UTC: Papermaster takes a moment to talk about superior CPU security. AMD microarchitectures largely immune to side-channel attacks. AMD implements memory encryption.


21:50 UTC: AMD has shipped 260 million+ Zen cores shipped to date!


21:51 UTC: AMD is committed to improving IPC of its processors. They're reporting on their single-threaded performance gains, and how they were achieved.


21:54 UTC: Papermaster now goes into more detail regarding future technologies. AMD's packaging innovations are a response to Moore's Law "slowing down". X3D packaging could be AMD's response to Intel Foveros. A step up from 3D packaging.


21:55 UTC: Bringing it all together is 3rd gen Infinity Fabric with more scalability and coherency. Also it has unified memory pools, starting with 3rd gen Infinity Fabric. What's interesting here is that AMD is no longer talking about "Infinity Fabric", but is now using the term "Infinity Architecture", to describe all silicon products that use IF technology in one way or another.


22:02 UTC: Culmination of all of AMD's scalability tech puts it ahead in the HPC competition.


22:05 UTC: Next up: Graphics. David Wang of Radeon Technology Group takes center stage and introduces their strategy looking forward. AMD is the first company to introduce domain-specific architectures: RDNA is for gaming, CDNA is for compute.


22:13 UTC: 50% performance improvement over RDNA with RDNA2


22:10 UTC: Details on RDNA now and tomorrow. RDNA2 has Variable Rate Shading and native support for hardware-accelerated ray-tracing.


22:14 UTC: All new RDNA2 based hardware-accelerated ray-tracing architecture, leverages DXR 1.1, which is a new revision of Microsoft's DirectX Raytracing API. The slide also shows a screenshot of raytracing running on AMD's RDNA2 silicon.


22:15 UTC: AMD's CDNA architecture is compute-optimized, removes all raster hardware. Adds tensor hardware to help AI. Highly scalable. Backed by open-source software, unlike NVIDIA. Existing CUDA code can be converted with HIP code converters.


22:24 UTC: AMD PC business overview.


22:30 UTC: AMD has big plans to grow its CPU market share in notebooks, across all form-factors.


22:33 UTC: AMD talks mobile security, Ryzen Pro.


22:37 UTC: A brief overview of growth in gaming.



22:39 UTC: Brief overview of Navi. Big push for variable-rate-shading (VRS), ray-tracing, and process improvements with RDNA2. AMD's upcoming Navi2X GPUs are described as "Top-of Stack GPUs", "with uncompromising 4K Gaming" — finally some competition for NVIDIA!


22:43 UTC: AMD chronicles the today and tomorrow of console gaming.


23:09 UTC: Session 2 resumes after a brief recess. Forrest Norrod of Data Center computing takes stage.


23:14 UTC: If you're impatient about Forrest Norrod's presentation, here are the slides.


23:20 UTC: Cloud providers love EPYC. The processor is powering Twitter and CloudFlare.


23:24 UTC: HPC is a key technological growth area for AMD on the back of huge design wins.


23:25 UTC: AMD is betting on telecom, particularly the 5G rollout, which has high requirements for IO connectivity — which EPYC is able to provide. 3rd Gen EPYC "Milan" processors will be shipping late-2020.


23:28 UTC: AMD's 4th generation EPYC processors are codenamed "Genoa", and powered by Zen 4 CPU cores. A 5 nm fabrication process is used, coming in 2022.


23:30 UTC: Data Center GPUs both for supporting virtual desktops/gaming, and AI/ML a big bet for AMD.


23:34 UTC: Data Centers are handling most of the computing. 10X performance growth needed over the next few years. AMD's Data Center GPU Roadmap shows CDNA, which is optimized for ML/HPC, and CDNA2, which extends the computing capabilities to Exascale. The CDNA2 architecture will also be cache-coherent with the main processor, which means that workloads can live in any memory and don't have to be copied around anymore.


23:41 UTC: Devinder Kumar, CFO takes stage to talk about the company's financials since 2017 and the way forward.



23:43 UTC: Devinder Kumar is presenting AMD financial details now, here's his full slide deck. Data Centers will make up nearly 1/3rd of AMD's revenues in the coming years.



23:55 UTC: CEO Lisa Su summarizes key takeaways before Q&A.



23:58 UTC: CEO Lisa Su talks about the Coronavirus: Protecting employees against COVID-19 at a policy level, supply chains largely unaffected, although partners dependent on China somewhat affected.

Q&A

00:07 UTC: RDNA and CDNA IPC scaling identical.

00:20 UTC: Zen 4 is 5 nm, but AMD is vague on RDNA3 and CDNA2 process. CDNA architecture physically lacks the raster bulk of RDNA, replaced with tensor accelerators (matrix multipliers).

00:31 UTC: And that's a wrap. We will dissect much of the information presented here shortly, into separate news posts.

View at TechPowerUp Main Site
 
Joined
Jun 3, 2010
Messages
2,540 (0.50/day)
Me likey, especially the part about foundry lead catch up, lol. It was super cheeky. :)

All current AMD line ups are works of art. All is well except their gpus and with their excessive voltage regulation. I say release Timothy Lottes and say, "Full power!" to rapid packed math scalarized code in all workloads. They have a good math library compiler now and AMD is specializing in the gpu semi-custom era. Why don't we have optimized low latency code for their gpu titles more often, rapid packed math ought to be the evolution of heterogeneous system architecture.
 
Last edited:

Aquinus

Resident Wat-man
Joined
Jan 28, 2012
Messages
13,147 (2.94/day)
Location
Concord, NH, USA
System Name Apollo
Processor Intel Core i9 9880H
Motherboard Some proprietary Apple thing.
Memory 64GB DDR4-2667
Video Card(s) AMD Radeon Pro 5600M, 8GB HBM2
Storage 1TB Apple NVMe, 4TB External
Display(s) Laptop @ 3072x1920 + 2x LG 5k Ultrafine TB3 displays
Case MacBook Pro (16", 2019)
Audio Device(s) AirPods Pro, Sennheiser HD 380s w/ FIIO Alpen 2, or Logitech 2.1 Speakers
Power Supply 96w Power Adapter
Mouse Logitech MX Master 3
Keyboard Logitech G915, GL Clicky
Software MacOS 12.1
Several months ago Jim Keller did a talk and he was talking about how 3D printing of circuits is (likely,) going to be the future, but it comes with all sorts of interesting issues that haven't had to be considered in the past, such as the heat flux through multiple layers of heat producing components. HBM has the same kind of issue, but DRAM traditionally hasn't been all that power hungry. I totally would love to see a good SoC that included a base level of system memory on the same package as the rest of the CPU. For setups where space is at a premium, like in laptops or very low profile builds, having everything on the same package could be a big bonus.
 
Joined
Mar 16, 2017
Messages
1,666 (0.64/day)
Location
Tanagra
System Name Budget Box
Processor Xeon E5-2667v2
Motherboard ASUS P9X79 Pro
Cooling Some cheap tower cooler, I dunno
Memory 32GB 1866-DDR3 ECC
Video Card(s) XFX RX 5600XT
Storage WD NVME 1GB
Display(s) ASUS Pro Art 27"
Case Antec P7 Neo
50% increase in performance per watt over RDNA is nice to hear, and that would really help the case that “Big Navi” is a possibility without blowing the power budget.
 
Joined
Jun 3, 2010
Messages
2,540 (0.50/day)
Several months ago Jim Keller did a talk and he was talking about how 3D printing of circuits is (likely,) going to be the future, but it comes with all sorts of interesting issues that haven't had to be considered in the past, such as the heat flux through multiple layers of heat producing components
I recently noticed an interesting unrelated fact - did you know that it was Lamborghini designing the Dodge Viper V10 engine?
They didn't want any launch scruples, so the original engine had extra weight for added safety - some 65kg, 301kg total, which got shed in later generations. Likewise, Intel needs a solid footing prior to reinventing the HBM wheel, imo.
 

Aquinus

Resident Wat-man
Joined
Jan 28, 2012
Messages
13,147 (2.94/day)
Location
Concord, NH, USA
System Name Apollo
Processor Intel Core i9 9880H
Motherboard Some proprietary Apple thing.
Memory 64GB DDR4-2667
Video Card(s) AMD Radeon Pro 5600M, 8GB HBM2
Storage 1TB Apple NVMe, 4TB External
Display(s) Laptop @ 3072x1920 + 2x LG 5k Ultrafine TB3 displays
Case MacBook Pro (16", 2019)
Audio Device(s) AirPods Pro, Sennheiser HD 380s w/ FIIO Alpen 2, or Logitech 2.1 Speakers
Power Supply 96w Power Adapter
Mouse Logitech MX Master 3
Keyboard Logitech G915, GL Clicky
Software MacOS 12.1
I recently noticed an interesting unrelated fact - did you know that it was Lamborghini designing the Dodge Viper V10 engine?
They didn't want any launch scruples, so the original engine had extra weight for added safety - some 65kg, 301kg total, which got shed in later generations. Likewise, Intel needs a solid footing prior to reinventing the HBM wheel, imo.
Not to go too far off topic, but I have a lotheing for V10 engines because they're inherently unbalanced. I like engines that don't have primary and secondary balance issues. (Not to mention I love the sound of the Flat-4 in my Impreza.) :laugh:

...but to stay on topic, I suspect that Intel is working on something. It's not like it's anything they haven't done before if you consider Crystalwell, but they haven't dedicated nearly as many resources into it as AMD has and it's finally paying for (for AMD.) Intel needs to do something to improve yields. They can't play this game forever.
 
Joined
Jul 19, 2016
Messages
476 (0.17/day)
Interesting that Zen 3 is showing under '7nm' and not '7nm+' as expected. Did they not want to go that fine-grained in the slide or did plans change?
 
Joined
Jun 3, 2010
Messages
2,540 (0.50/day)
Interesting that Zen 3 is showing under '7nm' and not '7nm+' as expected. Did they not want to go that fine-grained in the slide or did plans change?
Afaik, 7nm is directly implementable to 5nm. That said, however, AMD just purchased extra volume on left over TMSC 7nm lines. That would be why, imo.
 
Joined
Jul 29, 2019
Messages
77 (0.04/day)
Interesting that Zen 3 is showing under '7nm' and not '7nm+' as expected. Did they not want to go that fine-grained in the slide or did plans change?
Yeah, noticed that too. Did they change the plans, or did something change at TSMC?
 
Joined
Jun 3, 2010
Messages
2,540 (0.50/day)
Starting with 14nm that got adopted from Samsung, GlobalFoundries CTO that later became the CEO, made a daring exclamation that they wouldn't risk it process wise in a design first launch. Basically, tic-tock(I never thought would have said that).
AMD is keeping up with such sound process practices more than anything, don't read between the lines.
 
Joined
Jan 8, 2017
Messages
8,929 (3.36/day)
System Name Good enough
Processor AMD Ryzen R9 7900 - Alphacool Eisblock XPX Aurora Edge
Motherboard ASRock B650 Pro RS
Cooling 2x 360mm NexXxoS ST30 X-Flow, 1x 360mm NexXxoS ST30, 1x 240mm NexXxoS ST30
Memory 32GB - FURY Beast RGB 5600 Mhz
Video Card(s) Sapphire RX 7900 XT - Alphacool Eisblock Aurora
Storage 1x Kingston KC3000 1TB 1x Kingston A2000 1TB, 1x Samsung 850 EVO 250GB , 1x Samsung 860 EVO 500GB
Display(s) LG UltraGear 32GN650-B + 4K Samsung TV
Case Phanteks NV7
Power Supply GPS-750C
I suspect that Intel is working on something.

Well if that's the case then they're eerily silent about it. I mean it's extremely odd how for the past 2 years or so they've just sort of sat there not talking about a whole lot of upcoming stuff. I mean yes, they've shown a prototype for a GPU, Foveros and a new architecture for mobile but that's hardly substantial it's just no where on the scale of what AMD and Nvidia seem to be doing. As much as I dislike them I do genuinely wonder what the hell are they doing.

Yeah, noticed that too. Did they change the plans, or did something change at TSMC?

Milan is supposed to come out this year and 7nm+ is likely not ready for volume. Probably for the same reason it looks like Nvidia wont use TSMC.
 
Joined
Mar 16, 2017
Messages
1,666 (0.64/day)
Location
Tanagra
System Name Budget Box
Processor Xeon E5-2667v2
Motherboard ASUS P9X79 Pro
Cooling Some cheap tower cooler, I dunno
Memory 32GB 1866-DDR3 ECC
Video Card(s) XFX RX 5600XT
Storage WD NVME 1GB
Display(s) ASUS Pro Art 27"
Case Antec P7 Neo
Interesting that Zen 3 is showing under '7nm' and not '7nm+' as expected. Did they not want to go that fine-grained in the slide or did plans change?
Yeah, noticed that too. Did they change the plans, or did something change at TSMC?
Anandtech said that when AMD had put Zen3 on the roadmap and listed it as 7nm+, they meant it to be “an advanced 7nm.” Then TSMC announced a official process called “7nm+,” but they aren’t one in the same. Zen3 will still be an advancement, just not using the official 7nm+ process from TSMC.
 
Joined
Aug 8, 2019
Messages
430 (0.25/day)
System Name R2V2 *In Progress
Processor Ryzen 7 2700
Motherboard Asrock X570 Taichi
Cooling W2A... water to air
Memory G.Skill Trident Z3466 B-die
Video Card(s) Radeon VII repaired and resurrected
Storage Adata and Samsung NVME
Display(s) Samsung LCD
Case Some ThermalTake
Audio Device(s) Asus Strix RAID DLX upgraded op amps
Power Supply Seasonic Prime something or other
Software Windows 10 Pro x64
Give me a 400W GPU if it's got the performance, I don't care. Halo products exist for reasons.
 
Joined
Mar 23, 2005
Messages
4,061 (0.58/day)
Location
Ancient Greece, Acropolis (Time Lord)
System Name RiseZEN Gaming PC
Processor AMD Ryzen 7 5800X @ Auto
Motherboard Asus ROG Strix X570-E Gaming ATX Motherboard
Cooling Corsair H115i Elite Capellix AIO, 280mm Radiator, Dual RGB 140mm ML Series PWM Fans
Memory G.Skill TridentZ 64GB (4 x 16GB) DDR4 3200
Video Card(s) ASUS DUAL RX 6700 XT DUAL-RX6700XT-12G
Storage Corsair Force MP500 480GB M.2 & MP510 480GB M.2 - 2 x WD_BLACK 1TB SN850X NVMe 1TB
Display(s) ASUS ROG Strix 34” XG349C 180Hz 1440p + Asus ROG 27" MG278Q 144Hz WQHD 1440p
Case Corsair Obsidian Series 450D Gaming Case
Audio Device(s) SteelSeries 5Hv2 w/ Sound Blaster Z SE
Power Supply Corsair RM750x Power Supply
Mouse Razer Death-Adder + Viper 8K HZ Ambidextrous Gaming Mouse - Ergonomic Left Hand Edition
Keyboard Logitech G910 Orion Spectrum RGB Gaming Keyboard
Software Windows 11 Pro - 64-Bit Edition
Benchmark Scores I'm the Doctor, Doctor Who. The Definition of Gaming is PC Gaming...
50% increase in performance per watt over RDNA is nice to hear, and that would really help the case that “Big Navi” is a possibility without blowing the power budget.
Based on most of the rumours, RDNA2 is supposed to be a very efficient design in performance and in power draw.
 
Joined
Mar 23, 2005
Messages
4,061 (0.58/day)
Location
Ancient Greece, Acropolis (Time Lord)
System Name RiseZEN Gaming PC
Processor AMD Ryzen 7 5800X @ Auto
Motherboard Asus ROG Strix X570-E Gaming ATX Motherboard
Cooling Corsair H115i Elite Capellix AIO, 280mm Radiator, Dual RGB 140mm ML Series PWM Fans
Memory G.Skill TridentZ 64GB (4 x 16GB) DDR4 3200
Video Card(s) ASUS DUAL RX 6700 XT DUAL-RX6700XT-12G
Storage Corsair Force MP500 480GB M.2 & MP510 480GB M.2 - 2 x WD_BLACK 1TB SN850X NVMe 1TB
Display(s) ASUS ROG Strix 34” XG349C 180Hz 1440p + Asus ROG 27" MG278Q 144Hz WQHD 1440p
Case Corsair Obsidian Series 450D Gaming Case
Audio Device(s) SteelSeries 5Hv2 w/ Sound Blaster Z SE
Power Supply Corsair RM750x Power Supply
Mouse Razer Death-Adder + Viper 8K HZ Ambidextrous Gaming Mouse - Ergonomic Left Hand Edition
Keyboard Logitech G910 Orion Spectrum RGB Gaming Keyboard
Software Windows 11 Pro - 64-Bit Edition
Benchmark Scores I'm the Doctor, Doctor Who. The Definition of Gaming is PC Gaming...
PR slides FTW. Everyone will just get Ampere before RDNA2 even comes out lel
Ampere? Since when do gamers buy server GPUs lol
 
Joined
Nov 3, 2013
Messages
2,141 (0.56/day)
Location
Serbia
Processor Ryzen 3600
Motherboard X570 I Aorus Pro
Cooling Deepcool AG400
Memory HyperX Fury 2 x 8GB 3200 CL16
Video Card(s) RX 470 Nitro+ 4GB
Storage SX8200 Pro 512 / NV2 512
Display(s) 24G2U
Case NR200P
Power Supply Ion SFX 650
Mouse G703
Keyboard Keychron V1 (Akko Matcha Green) / Apex m500 (gateron milky yellow)
Software W10
That 50% perf per watt on RDNA2 looks juicy, but late 2020 might be too late... again.


EDIT:
Thank god
blower.png
 
Joined
Nov 11, 2016
Messages
3,065 (1.13/day)
System Name The de-ploughminator Mk-II
Processor i7 13700KF
Motherboard MSI Z790 Carbon
Cooling ID-Cooling SE-226-XT + Phanteks T30
Memory 2x16GB G.Skill DDR5 7200Cas34
Video Card(s) Asus RTX4090 TUF
Storage Kingston KC3000 2TB NVME
Display(s) LG OLED CX48"
Case Corsair 5000D Air
Power Supply Corsair HX850
Mouse Razor Viper Ultimate
Keyboard Corsair K75
Software win11
That 50% perf per watt on RDNA2 looks juicy, but late 2020 might be too late... again.

Yeah by 50% efficiency gain AMD probably compare it to the 5700XT, which means RDNA2 (at proper voltage) is about 40% higher effieciency than Turing. A die shrunk Turing at 7nm+ should be 75% more efficient without any architectural change. The only thing AMD can do is wait for Nvidia to show their hands then do some last minute adjustments to clocks/prices to outmatch Nvidia mid-ranges a little and everyone is happy.
 
Joined
Apr 12, 2013
Messages
6,749 (1.68/day)
A die shrunk Turing at 7nm+ should be 75% more efficient without any architectural change.
And where did you get that number from?
The only thing AMD can do is wait for Nvidia
Yeah sure, we've heard that before especially with the CPUs :rolleyes:

While I don't expect AMD to overtake Nvidia in the near future, the comments these days will make you feel RDNA (series) is the dozer to Intel's (Nvidia) Sandy Bridge ~ which is so far away from the truth it's not even funny o_O

I'll paste these here again, for people who have difficulty understanding how efficiency works at different levels & even GPU classes ~
 
Joined
Jul 9, 2015
Messages
3,413 (1.06/day)
System Name M3401 notebook
Processor 5600H
Motherboard NA
Memory 16GB
Video Card(s) 3050
Storage 500GB SSD
Display(s) 14" OLED screen of the laptop
Software Windows 10
Benchmark Scores 3050 scores good 15-20% lower than average, despite ASUS's claims that it has uber cooling.
Afaik, 7nm is directly implementable to 5nm. That said, however, AMD just purchased extra volume on left over TMSC 7nm lines. That would be why, imo.

7N and 7NP and onwards are design compatible.
7N+ (EUV) is not.

I don't expect AMD to overtake Nvidia in the near future
5600, 5700 series "overtook" nvidia.
As for that funny Ti thing for $1.3k, less than one 1% of the market buys them, why should it even matter to the remaining 99%?
 
Joined
Nov 11, 2016
Messages
3,065 (1.13/day)
System Name The de-ploughminator Mk-II
Processor i7 13700KF
Motherboard MSI Z790 Carbon
Cooling ID-Cooling SE-226-XT + Phanteks T30
Memory 2x16GB G.Skill DDR5 7200Cas34
Video Card(s) Asus RTX4090 TUF
Storage Kingston KC3000 2TB NVME
Display(s) LG OLED CX48"
Case Corsair 5000D Air
Power Supply Corsair HX850
Mouse Razor Viper Ultimate
Keyboard Corsair K75
Software win11
And where did you get that number from?
Yeah sure, we've heard that before especially with the CPUs :rolleyes:

While I don't expect AMD to overtake Nvidia in the near future, the comments these days will make you feel RDNA (series) is the dozer to Intel's (Nvidia) Sandy Bridge ~ which is so far away from the truth it's not even funny o_O

I'll paste these here again, for people who have difficulty understanding how efficiency works at different levels & even GPU classes

Yeah kinda hard for you to notice the 5700XT in the chart eh :D, it's like looking through color glasses only. FYI lower voltage increase efficiency, which 5700 and 5600XT demonstrate with maximum Vcore of 1.0v, but of course what AMD means by RDNA is always the 5700XT, not 5700 nor 5600XT. You just don't take a cut down version of a chip as a representative of an architecture.

Taken from https://www.techcenturion.com/7nm-10nm-14nm-fabrication
TSMC 7nm vs 12/16nm: 20% performance and 40% power reduction = 60% efficiency
TSMC 7nm+ vs 7nm: 10% performance or 15% power reduction = 15% efficiency
That explains the 75% efficiency increase for you ?

BTW Manufacturer OC cards also increase effiency, it's nothing new
 
Last edited:
Joined
Jul 19, 2016
Messages
476 (0.17/day)
Yeah kinda hard for you to notice the 5700XT in the chart eh :D, it's like looking through color glasses only. FYI lower voltage increase efficiency, which 5700 and 5600XT demonstrate with maximum Vcore of 1.0v, but of course what AMD mean RDNA is always the 5700XT, not 5700 nor 5600XT

Taken from https://www.techcenturion.com/7nm-10nm-14nm-fabrication
TSMC 7nm vs 12/16nm: 20% performance and 40% power reduction = 60% efficiency
TSMC 7nm+ vs 7nm: 10% performance or 15% power reduction = 15% efficiency
That explains the 75% efficiency increase for you ?

This is not how anything works.

In any case, even if performance per watt improvement is 75%, actual total performance gain is dependant on clockspeeds. I don't think they'll match clockspeeds on their first 7nm products compared to their 12nm cards.
 
Joined
Nov 11, 2016
Messages
3,065 (1.13/day)
System Name The de-ploughminator Mk-II
Processor i7 13700KF
Motherboard MSI Z790 Carbon
Cooling ID-Cooling SE-226-XT + Phanteks T30
Memory 2x16GB G.Skill DDR5 7200Cas34
Video Card(s) Asus RTX4090 TUF
Storage Kingston KC3000 2TB NVME
Display(s) LG OLED CX48"
Case Corsair 5000D Air
Power Supply Corsair HX850
Mouse Razor Viper Ultimate
Keyboard Corsair K75
Software win11
This is not how anything works.

In any case, even if performance per watt improvement is 75%, actual total performance gain is dependant on clockspeeds. I don't think they'll match clockspeeds on their first 7nm products compared to their 12nm cards.

Yeah sure.
Maxwell to Pascal (similar in architecture)
28nm vs 16nm which TSMC said bring almost 50% more performance and 60% more efficient
980 Ti 250W TDP 6.06 TFLOPS
1080 Ti 250W TDP 11.34 TFLOPS
That is 87% efficiency gain, yeah it's less than 110% promised by TSMC but it's almost. 1080Ti is also 85% faster than 980Ti in 4K benchmark.

Ampere is rumored to be 50% faster than Turing (3080 Ti vs 2080 Ti) at 50% power (most likely testing at reduced power limit), I reckon 3080Ti will end up 75% faster than 2080Ti at 100% power limit (260W TDP). Just recently the Geekbench leaks for Ampere already indicate that anyways.
 
Last edited:
Top