AMD's Future Chips & SoC's: News, Info & Rumours.

Page 49 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.


Precision Boost 2 was already present in first gen RyZen products. Moreover it is not any change to the microarchitecture but only a change in the management of boosts.

http://www.amd.com/en/technologies/sense-mi

Trinity also introduced resonant clock mesh to manage better clocks, but the core was the same Piledriver core than for other chips.

Optimized speed and latency for the memory controller surely means higher clocked stock configuration. But the Memory controller is not part of the core, neither the microarchitecture. So again it is all Zen cores.

Optimized speed and latency for cache is more of the same.
 


By that definition, how can Intel call anything since Skylake a different name? I mean, they are literally the same bloody chip, on the same process. The only difference is changing MB support and a more mature process allowing higher clocks.
 


Agree 100% with this i mean Intel has basically made no improvements since then in their architecture simply pulling an Amd by adding moar cores.
 


In the first place Intel uses a well-known tick-tock-optimization procedure. That is why there is Sandybridge and IvyBridge, then Haswell and Broadwell, and now Skylake, Kabylake and CoffeLake...

In the second place those codenames also refer to the GPU part of the CPU. For instance Kabylake includes a new graphics architecture compared to Skylake. It is not the same "bloody chip".

When Intel releases a chip is exactly the same chip but with higher clocks, Intel uses the term refresh as in "Haswell Refresh" or in Kabylake R (the R is for Refresh).
 
CES 2018 LIVESTREAM
http://live.ces.tech/

CES 2018, day 1: Live from the CNET Stage in Las Vegas
The doors are open on the world's biggest tech trade show and CNET is live from the show floor.
https://www.cnet.com/news/ces-2018-day-1-watch-the-cnet-stage-livestream/
 
Cloud companies consider Intel rivals after the discovery of microchip security flaws
Some of Intel's data center customers are exploring using microchips from its rivals to build new infrastructure after the discovery of security flaws affecting most chips.
Intel chips back 98 percent of data center operations, according to industry consultancy IDC.
Security researchers last week disclosed flaws, dubbed Meltdown and Spectre.
Published 33 Mins Ago

https://www.cnbc.com/2018/01/10/cloud-companies-consider-intel-rivals-after-security-flaws-found.html
Microsoft said on Tuesday the patches necessary to secure the threats could have a significant performance impact on servers.

Intel will help customers find the best approach in terms of security, performance, and compatibility, it said in a statement on Tuesday. "For many customers, the performance element is foremost, and we are sharply focused on doing all we can to ensure that we meet their expectations."

Alternatives include Advanced Micro Devices, which shares with Intel a chip architecture called x86, or chips based on technology from ARM Holdings or graphics processing chips, which were developed for different tasks than Intel and AMD's central processing units, or CPUs.

For Gleb Budman's company, San Mateo-based online storage firm Backblaze, building with ARM chips would not be difficult.

"If ARM provides enough computing power at lower cost or lower power than x86, it would be a strong incentive for us to switch," said Budman. "If the fix for x86 results in a dramatically decreased level of performance, that might increasingly push in favor of switching to ARM."

Infinitely Virtual, a Los Angeles-based cloud computing vendor, is counting on Intel to replace equipment or offer a rebate to make up for the loss in computing power, Chief Executive Adam Stern said in an interview.

"If Intel doesn't step up and do something to make this right then we're going to have to punish them in the marketplace by not purchasing their products," said Stern, whose company relies exclusively on Intel processors.


Cloud providers said swapping out previously installed Intel chips for rivals' would be too complex, but moving forward they could expand their networks using alternatives. Moving from Intel to AMD is easiest since AMD and Intel chips share a common core technology called the x86 instruction set, they said.

ARM-based chips lag the speed of Intel's x86 based chips for tasks such as searches, and software would have to be rewritten.

This could make for a potential push to AMD.
 
Microsoft halts some AMD chip patches after PCs suffer 'blue screen errors'
Microsoft said it had suspended patches to guard against Meltdown and Spectre security threats for computers running AMD chipsets.
Microsoft received complaints by AMD customers that the software updates froze their machines.
The software giant said the patches had caused computer screens to freeze up, or have so-called "blue screen errors" on Windows 10, Windows 8.1 and Windows 7.
https://www.cnbc.com/2018/01/09/microsoft-halts-some-amd-chip-meltdown-patches-after-pcs-freeze.html
http://www.eweek.com/security/microsoft-investigating-problematic-meltdown-patch-on-amd-pcs
https://arstechnica.com/gadgets/2018/01/bad-docs-and-blue-screens-make-microsoft-suspend-spectre-patch-for-amd-machines/
 


What does this have to do with Naples?

EDIT: I never notice the thread name change, lol. Well, question is still valid in the context of "future CPUs".
 


AMD's Future Chips & SoC's: News, Info & Rumours.....

They are news right?
😉
 


Actually,
Randomizer said:
Given that this discussion is taking place over at least three different threads I think we need a thread dedicated to discussing the security flaws so that the existing threads can remain focused on their original topics.
http://www.tomshardware.com/forum/id-3341285/amd-naples-server-cpu-info-rumours/page-24.html#20555768
so I started this thread
http://www.tomshardware.com/forum/id-3609004/cpu-security-vulnerabilities-information.html
Which I mentioned this same problem 2 days ago after one of our members/moderators darkbreeze mentioned the windows update was bricking his AMD system.
http://www.tomshardware.com/forum/id-3609004/cpu-security-vulnerabilities-information/page-4.html#20573695

The update is also affecting more than just AMD system I'm going to make a post about that now. Apparently, astronomy components aren't working right after this update as well.

Edit: Randomizer didn't want the same post across multiple threads as the reason for the sticky of the Meltdown & Spector thread.

2nd Edit: Yuka is right actually. AMD's Future Chips & SoC's: News, Info & Rumours. Old chips news doesn't apply to chips made in the future. Or the title would be just AMD Chips & SoC's.
 


Are you really asking that question? Really?
 


Like I said on my other thread, these people have no idea what they are talking about, That makes no sense at all, why should we do that?.. What's the point on changing over to AMD. The flaws affect all modern processors not only Intel (INTC), but AMD and ARM that use "speculative execution" to enhance performance. Why should we look to AMD or ARM???. Nearly all modern chip architectures from the major suppliers (Intel, AMD, ARM) are affected, and this includes nearly all modern computer systems from data center to PC to smartphones. The problem affects nearly all operating systems, such as Windows, Linux, macOS and even Android, as well as virtualized environments such as VMware and Citrix
 
That will really depend on what specific fix causes these performance losses. Meltdown has no need to be fixed on AMD CPUs, so if that part of the patch is responsible, their reaction would be absolutely legitimate.
 


Fair point, but the performance loss on the Desktop on the latest CPUs is basically just noise, so the price/performance question hasn't changed much. Yes, servers get hit, VMs get hit, and older CPUs (rip my 2600k) get hit. But for Skylake and later on Windows 10, the net performance loss for desktop users is basically zero.

I also have a suspicion there's a few attack avenues on modern CPUs that haven't been discovered yet. CPUs are now very complex beasts that have essentially co-CPUs on board, and I see a lot of potential avenues for attack that really haven't been that thoroughly explored yet. I really think its time all parties have a sitdown and figure out how to re-design their HW/SW with a focus on security. Can't help but wonder how OS/2's use of Ring 2 would help in these particular cases...maybe MSFT should dust off those old OS/2 design documents when work starts on W11?
 
Honestly, more testing needs to be done. For instance DigitalFoundry found Witcher 3 actually takes a ~10% performance hit. The Witcher 3, Ultra, No HairWorks pre-patch 139.8fps windows update 128.3fps windows+bios 126.6fps
https://youtu.be/LC1WuKdPVCQ
Also, techspot.com showed a 0-4% decrease in FPS on average. And up to 41% performance loss in some SSD and NVME synthetic test.
https://www.techspot.com/article/1556-meltdown-and-spectre-cpu-performance-windows/
Yuka has mentioned that a test needs to be done on streaming. I agree this likely cause another hit to performance for Intel. Plus, VR needs to be tested as well. All in all it's too early for definitives, and much more testing needs to be done across multiple reviewers.
 
With rumors stating the 2600 will have a 200 mhz jump its sounding like that is what we can expect from Zen+ and rumors our stating better memory overclocks which is exactly what we wanted.

Also is Amd killing the APU name?
 
Since they have unified the socket (platform), maybe they realized the segmentation was not needed anymore. Specially with Intel using the final letter as discriminator. I would imagine they'll just follow suit.

And from that video. I disagree slightly with 2 points:

1.- The IPC increase they might (big IF) get due to cache shenanigans and improves IMC (hard to accept they worked magic here) will be past the 5% mark. That being said, if (again) they do get a 5% increase just due to cache, it would be amazing. Then you add the clocks and you have a really solid product to contend against Covfefe Lake refreshes/updates. I've already seen Intel filling the rest of the lineup with 100Mhz boosts. That looks so lame... Not that AMD hasn't done it, but still...

2.- I don't think nVidia will sit on Pascal "just because they can". It doesn't make business sense to extend the life of Pascal more than what's already out there now. They have no certainty how the market will behave, so betting on current prices will continue to be jacked up and demand as it is now, they're putting a ton of risk I don't think is wise. They'll probably skip a cycle, but they will churn out Volta or Ampere when they're ready. Plus, from a pure selling perspective, if Volta and Ampere are better at mining than AMD, why would you lose the opportunity to overtake AMD there as well? I mean, it's money, right?

Cheers! 😛
 


A ~10% performance improvement means very little when your competitor is already 30--40% ahead in performance...
 


Pulling numbers from a dark place again, I see.

For all that I'd love for you to go hunt for a bazillion graphs that prove your point, I don't want to go look for the graphs that contradict it. It's just tiring to do.

Now, whatever you say, if they do indeed get a ~10% increase in brute performance due to (base) clock and other improvements that don't really alter Zen's inner shenanigans, it's a great gain given effort put. I'm pretty sure we'll see a decent bump for Zen v2 when it lands, so this refresh will be good to have until then. Like I already said, Intel is already filling up the line up in anticipation, so they're wary.
 
Status
Not open for further replies.