Intel's Future Chips: News, Rumours & Reviews

[logainofhades]

The intel i7 7800X is now only $309.. 😉 and compared with the Ryzen 7($299) offers better value and is faster. Is easy to tell witch technology is better when a 6 core beats an 8 core in overall performance in both single and multi threaded applications (without been overclocked)

i7 7800X single core score: 2250, multi core score: 14750

Ryzen 7 single core score 1771, multi core score 13756

Compare CPU here https://www.cpubenchmark.net/compare.php?cmp

Price/performance still isn't very good though, due to high motherboard cost, and quad channel ram. Also Ryzen 7, especially the 1700, doesn't need expensive cooling. Skylake-X are mini space heaters in comparison.You could easily buy a 500gb 960 Evo, for that price difference, or even a better GPU than would be possible for the same amount of money going with such a platform.

 

[goldstone77]

Also, in 7 years of so from now when Intel releases 10nm for desktops you are going to want to buy the newest thing! Don't spend a lot of money on old technology. Especially, with the rise of all these security flaws.

 

[logainofhades]

I am not overly concerned about the security flaws. I do concern myself with price/performance. When I bought my 6700k, it fit my needs in that area quite well. FX was simply not an option. Just so happens AMD finally has a worthwhile product, that if I had to build from scratch today, that I would consider using Ryzen.

 

[-Fran-]

I am not overly concerned about the security flaws. I do concern myself with price/performance. When I bought my 6700k, it fit my needs in that area quite well. FX was simply not an option. Just so happens AMD finally has a worthwhile product, that if I had to build from scratch today, that I would consider using Ryzen.

PREACH IT!

Well, I already put my new laptop review in the lappies thread, so I won't say anything here, but you don't need the "latest and greatest" to be a very satisfied customer anymore... For better or for worse, I think. Although it does sound kinda sad, haha.

Cheers!

 

[8350rocks]

The intel i7 7800X is now only $309.. 😉 and compared with the Ryzen 7($299) offers better value and is faster. Is easy to tell witch technology is better when a 6 core beats an 8 core in overall performance in both single and multi threaded applications (without been overclocked)

i7 7800X single core score: 2250, multi core score: 14750

Ryzen 7 single core score 1771, multi core score 13756

Compare CPU here https://www.cpubenchmark.net/compare.php?cmp

A couple things to note: Most reviewers do not use cpubenchmark

http://cpu.userbenchmark.com/Compare/Intel-Core-i7-7800X-vs-AMD-Ryzen-7-1700X/m304816vs3915
has some interesting comparisons to make. If on a limited budget you can put a 1700X on a $50-$100 motherboard using the rest on better components.

Most reviewers are full of it...I mostly use cpubenchmark and userbenchmark,,I know Ryzen 7 has a higher market share because it was cheaper(that's why I bought one for my wife)but now the core i7 7800X beats Ryzen in price vs performance so soon that gap may change. but x299 is not for everyone(always been intel's exclusive for the high end desktop), is not even a direct competitor of AM4. Anyway I don't want a cheap motherboard or a cheap rig, and not everyone wants a cheap motherboard. X299 has more features than am4. My pc case has glass panels with a liquid cooling loop and RGB fans. I don't want my motherboard to look like a black piece of cardboard plus I need RGB support and water pump support, be good at overclocking, I like on board wifi, good gaming audio and software with a good headphone amplifier and great ASUS bios for overclocking, intel optane memory support, quad channel memory, that being said am4 is not for me

Got my i7 7800X rock solid at 5.0GHz and is doing circles around my wife's Ryzen 7 gaming rig at rendering, gaming and at everything else. 😉, single core speed is not even comparable.
People don't know what they are talking about most programs and games will not benefit from 16 cores all those synthetic benchmarks are useless in real world desktop applications , thats why im not using a 36 core xeon. Rather stick with strong IPC and high speed per core.. I rather pay a bit more and have a better system..
Same that when I buy a phone I rather pay more and get a Samsung galaxy or pay more and get a faster better looking sport car or pay more and get a better looking house. That's what we do, we always want to have the best we could get.
Comparing Intel vs AMD. (intel could be more expensive)They are like a house and Both could have 4 bedrooms but in the master bedroom intel has a jacuzzi with a beer and wine cooler next to it

No worries, an upcoming windows patch will "fix" that for you...

How are those AMD patches working out for you? Not good if you own a older AMD CPU..

Microsoft has suspended delivering the latest Windows update to certain systems with AMD processors after reports that the update was causing the machines to crash with a blue screen of death when booting. The update contains countermeasures against both the Meltdown and Spectre attacks; although AMD systems are not affected by Meltdown, they're vulnerable to Spectre.
It's not entirely clear what the documentation error is. There are some reports that users with older AMD processors are seeing crashes because the Windows kernel is trying to use an instruction that the very earliest 64-bit AMD chips didn't support. It's not clear if this is the cause of suspension, but the bulk of the complaints do seem to concern older chips, so it's certainly possible.

Official statement from Microsoft:::
"After investigating, Microsoft has determined that some AMD chipsets do not conform to the documentation previously provided to Microsoft to develop the Windows operating system mitigations to protect against the chipset vulnerabilities known as Spectre and Meltdown."

https://www.cnbc.com/2018/01/09/microsoft-halts-some-amd-chip-meltdown-patches-after-pcs-freeze.html
https://arstechnica.com/information-technology/2014/08/after-blue-screen-of-death-reports-microsoft-says-to-uninstall-recent-patch/
https://arstechnica.com/gadgets/2018/01/bad-docs-and-blue-screens-make-microsoft-suspend-spectre-patch-for-amd-machines/

Zero impact on construction cores or newer...thanks for asking.

 

[8350rocks]

True but still far from those 35% (up to 50%) some news sites were pretending a week ago.

Also the BIOS updates are for Spectre, and as Linus mentioned those first updates are emergency fixes with no optimization for performance

https://www.realworldtech.com/forum/?threadid=173959&curpostid=173981

Can you please stop getting % numbers from your behind? You started complaining with "30%" claims at the very start and we all told you they were wrong for desktops due to the nature of the fixes. We're trying to look at facts here and you keep mixing up reports from credible sources with your own imagination and dubious reading.

Sure, there are people and sites misleading fellas, but WE ALREADY KNOW THAT. No one here is oblivious enough to claim desktops will see a massive drop in performance and the asterisk has always been around "mixed workloads". Until more sites actually review mixed workloads and not single application situations, then you won't really see anything wrong. This will be more of a subjective measurement, since you can't really tell with specific bench numbers how good is your system at handling mixed workloads without profiling the whole machine.

I'll keep an eye on Toms report for the Streaming piece (if they ever decide to update the figures). I have the suspicion Intel will get hit in that arena... At least enough to make it noticeable.

I have several contacts in financial datacenter IT arena, and it is looking bad for them. One of them works for a large banking chain in the U.S. and has supposed that they will need approximately 25% more hardware to accommodate normal workloads after the patch based on current internal projections. Willing to guess what he is going to recommend they buy to supplement existing hardware? I will give you 3 guesses, but I am betting you will only need one.

 

[YoAndy]

The intel i7 7800X is now only $309.. 😉 and compared with the Ryzen 7($299) offers better value and is faster. Is easy to tell witch technology is better when a 6 core beats an 8 core in overall performance in both single and multi threaded applications (without been overclocked)

i7 7800X single core score: 2250, multi core score: 14750

Ryzen 7 single core score 1771, multi core score 13756

Compare CPU here https://www.cpubenchmark.net/compare.php?cmp

Price/performance still isn't very good though, due to high motherboard cost, and quad channel ram. Also Ryzen 7, especially the 1700, doesn't need expensive cooling. Skylake-X are mini space heaters in comparison.You could easily buy a 500gb 960 Evo, for that price difference, or even a better GPU than would be possible for the same amount of money going with such a platform.

For some yes,(like new builders) But I already had a custom liquid loop. Sometimes I do prefer to spend a bit more to get a better future upgrade path or better gaming performance., or just because I want a little better, I do have a Ryzen 7, is my wife's gaming rig and I wouldn't trade my 7800X for it.

 

[aldaia]

I have several contacts in financial datacenter IT arena, and it is looking bad for them. One of them works for a large banking chain in the U.S. and has supposed that they will need approximately 25% more hardware to accommodate normal workloads after the patch based on current internal projections. Willing to guess what he is going to recommend they buy to supplement existing hardware? I will give you 3 guesses, but I am betting you will only need one.

WOW. That is really terrible consequences. I wonder if all companies incurring in such high costs will consider suing the hardware/OS makers. After all some of those companies are much bigger and powerful than intel (not to mention AMD).

Guess 1 :AMD ?

 

[juanrga]

I have several contacts in financial datacenter IT arena, and it is looking bad for them. One of them works for a large banking chain in the U.S. and has supposed that they will need approximately 25% more hardware to accommodate normal workloads after the patch based on current internal projections. Willing to guess what he is going to recommend they buy to supplement existing hardware? I will give you 3 guesses, but I am betting you will only need one.

WOW. That is really terrible consequences. I wonder if all companies incurring in such high costs will consider suing the hardware/OS makers. After all some of those companies are much bigger and powerful than intel (not to mention AMD).

Guess 1 :AMD ?

Remember AMD systems are also patched:

"In general, our experience is that Variant 1 and Variant 3 mitigations have minimal performance impact, while Variant 2 remediation, including OS and microcode, has a performance impact," Microsoft added.

Variants 1 and 3 refer to Spectre and Meltdown vulnerabilities, respectively, while Variant 2 is also Spectre. With that being the case, we suspect there will be a performance penalty on AMD systems too, since AMD is affected by Spectre.

 

[goldstone77]

Spoiler

I have several contacts in financial datacenter IT arena, and it is looking bad for them. One of them works for a large banking chain in the U.S. and has supposed that they will need approximately 25% more hardware to accommodate normal workloads after the patch based on current internal projections. Willing to guess what he is going to recommend they buy to supplement existing hardware? I will give you 3 guesses, but I am betting you will only need one.

WOW. That is really terrible consequences. I wonder if all companies incurring in such high costs will consider suing the hardware/OS makers. After all some of those companies are much bigger and powerful than intel (not to mention AMD).

Guess 1 :AMD ?

Remember AMD systems are also patched:

"In general, our experience is that Variant 1 and Variant 3 mitigations have minimal performance impact, while Variant 2 remediation, including OS and microcode, has a performance impact," Microsoft added.

Variants 1 and 3 refer to Spectre and Meltdown vulnerabilities, respectively, while Variant 2 is also Spectre. With that being the case, we suspect there will be a performance penalty on AMD systems too, since AMD is affected by Spectre.

Actually, the patch for Meltdown is the real kicker for Intel. It's causing 7 times as many system calls per request. That overhead is what's killing Intel. AMD is actually the least affected. Meltdown doesn't affect AMD. Spectre variant 2 AMD doesn't think they will be subject to it, and it hasn't been exploited on AMD systems. Spectre variant 1 is mitigated through OS patch only requires 1 extra instruction to patch, so there won't be a big performance hit on AMD just Intel.

Edit: Fixed errors

 

[juanrga]

Spoiler

I have several contacts in financial datacenter IT arena, and it is looking bad for them. One of them works for a large banking chain in the U.S. and has supposed that they will need approximately 25% more hardware to accommodate normal workloads after the patch based on current internal projections. Willing to guess what he is going to recommend they buy to supplement existing hardware? I will give you 3 guesses, but I am betting you will only need one.

WOW. That is really terrible consequences. I wonder if all companies incurring in such high costs will consider suing the hardware/OS makers. After all some of those companies are much bigger and powerful than intel (not to mention AMD).

Guess 1 :AMD ?

Remember AMD systems are also patched:

"In general, our experience is that Variant 1 and Variant 3 mitigations have minimal performance impact, while Variant 2 remediation, including OS and microcode, has a performance impact," Microsoft added.

Variants 1 and 3 refer to Spectre and Meltdown vulnerabilities, respectively, while Variant 2 is also Spectre. With that being the case, we suspect there will be a performance penalty on AMD systems too, since AMD is affected by Spectre.

Actually, the patch for Meltdown is the real kicker for Intel. It's causing 7 times as many system calls per request. That overhead is what's killing Intel. AMD is actually the least affected. Meltdown doesn't affect AMD. Spectre variant 1 is mitigated through OS patch, and variant 2 only requires 1 extra instruction to patch, so there won't be a big performance hit on AMD just Intel.

As Microsoft mentions above it is the Spectre patch which produces a noticeable performance impact, not the Meltdown patch. This is confirmed by third party reviews that tested Meltdown vs Meltdown+Spectre.

 

[-Fran-]

Spoiler

I have several contacts in financial datacenter IT arena, and it is looking bad for them. One of them works for a large banking chain in the U.S. and has supposed that they will need approximately 25% more hardware to accommodate normal workloads after the patch based on current internal projections. Willing to guess what he is going to recommend they buy to supplement existing hardware? I will give you 3 guesses, but I am betting you will only need one.

WOW. That is really terrible consequences. I wonder if all companies incurring in such high costs will consider suing the hardware/OS makers. After all some of those companies are much bigger and powerful than intel (not to mention AMD).

Guess 1 :AMD ?

Remember AMD systems are also patched:

"In general, our experience is that Variant 1 and Variant 3 mitigations have minimal performance impact, while Variant 2 remediation, including OS and microcode, has a performance impact," Microsoft added.

Variants 1 and 3 refer to Spectre and Meltdown vulnerabilities, respectively, while Variant 2 is also Spectre. With that being the case, we suspect there will be a performance penalty on AMD systems too, since AMD is affected by Spectre.

Actually, the patch for Meltdown is the real kicker for Intel. It's causing 7 times as many system calls per request. That overhead is what's killing Intel. AMD is actually the least affected. Meltdown doesn't affect AMD. Spectre variant 1 is mitigated through OS patch, and variant 2 only requires 1 extra instruction to patch, so there won't be a big performance hit on AMD just Intel.

As Microsoft mentions above it is the Spectre patch which produces a noticeable performance impact, not the Meltdown patch. This is confirmed by third party reviews that tested Meltdown vs Meltdown+Spectre.

The context of the discussion was Servers. Are those tests done in Server environments?

 

[goldstone77]

Spoiler

I have several contacts in financial datacenter IT arena, and it is looking bad for them. One of them works for a large banking chain in the U.S. and has supposed that they will need approximately 25% more hardware to accommodate normal workloads after the patch based on current internal projections. Willing to guess what he is going to recommend they buy to supplement existing hardware? I will give you 3 guesses, but I am betting you will only need one.

WOW. That is really terrible consequences. I wonder if all companies incurring in such high costs will consider suing the hardware/OS makers. After all some of those companies are much bigger and powerful than intel (not to mention AMD).

Guess 1 :AMD ?

Remember AMD systems are also patched:

"In general, our experience is that Variant 1 and Variant 3 mitigations have minimal performance impact, while Variant 2 remediation, including OS and microcode, has a performance impact," Microsoft added.

Variants 1 and 3 refer to Spectre and Meltdown vulnerabilities, respectively, while Variant 2 is also Spectre. With that being the case, we suspect there will be a performance penalty on AMD systems too, since AMD is affected by Spectre.

Actually, the patch for Meltdown is the real kicker for Intel. It's causing 7 times as many system calls per request. That overhead is what's killing Intel. AMD is actually the least affected. Meltdown doesn't affect AMD. Spectre variant 1 is mitigated through OS patch, and variant 2 only requires 1 extra instruction to patch, so there won't be a big performance hit on AMD just Intel.

As Microsoft mentions above it is the Spectre patch which produces a noticeable performance impact, not the Meltdown patch. This is confirmed by third party reviews that tested Meltdown vs Meltdown+Spectre.

As I mentioned AMD will not be impacted by these patches in a meaningful way, but Intel is affect to a greater extent!
Time stamp: https://youtu.be/Q7pVcYMobqc?t=372

 

[juanrga]

The context of the discussion was Servers. Are those tests done in Server environments?

https://cloudblogs.microsoft.com/microsoftsecure/2018/01/09/understanding-the-performance-impact-of-spectre-and-meltdown-mitigations-on-windows-systems/

 

[goldstone77]

The context of the discussion was Servers. Are those tests done in Server environments?

https://cloudblogs.microsoft.com/microsoftsecure/2018/01/09/understanding-the-performance-impact-of-spectre-and-meltdown-mitigations-on-windows-systems/

Even if what they claim is true, which their are reports to the contrary with just Meltdown patches on linux systems(1 system call mapped into 7). Variant 2 patching causes a further overhead for Intel systems on linux systems. AMD claims that Variant 2 hasn't been shown to affect AMD systems, so they are not patching for it in the microcode. So, we are back to AMD not being affected again.

Variant 1 and Variant 3 mitigations have minimal performance impact, while Variant 2 remediation, including OS and microcode, has a performance impact.

https://www.amd.com/en/corporate/speculative-execution

Differences in AMD architecture mean there is a near zero risk of exploitation of this variant. Vulnerability to Variant 2 has not been demonstrated on AMD processors to date.

 

[juanrga]

The context of the discussion was Servers. Are those tests done in Server environments?

https://cloudblogs.microsoft.com/microsoftsecure/2018/01/09/understanding-the-performance-impact-of-spectre-and-meltdown-mitigations-on-windows-systems/

Even if what they claim is true, which their are reports to the contrary with just Meltdown patches on linux systems(1 system call mapped into 7). Variant 2 patching causes a further overhead for Intel systems on linux systems. AMD claims that Variant 2 hasn't been shown to affect AMD systems, so they are not patching for it in the microcode. So, we are back to AMD not being affected again.

Variant 1 and Variant 3 mitigations have minimal performance impact, while Variant 2 remediation, including OS and microcode, has a performance impact.

https://www.amd.com/en/corporate/speculative-execution

Differences in AMD architecture mean there is a near zero risk of exploitation of this variant. Vulnerability to Variant 2 has not been demonstrated on AMD processors to date.

Several reports in the Internet reporting supposed Meltdown-patches performance are in reality measuring Spectre-patch impact. E.g. Amazon slowdown is due to Spectre patches, not Meltdown

https://www.realworldtech.com/forum/?threadid=173959&curpostid=173981

And now we have Microsoft confirming that the real problem is Spectre. This is understandable because the 'fix' for Variant 2 of the vulnerability consists on disabling branch prediction, which affects performance.

I know AMD PR claims that their CPUs are safe... but AMD engineers are working with kernel developers to patch the CPUs for Variant 2

https://lists.opensuse.org/opensuse-security-announce/2018-01/msg00004.html
https://www.suse.com/es-es/support/update/announcement/2018/suse-su-20180011-1/
https://bugs.gentoo.org/643476
https://access.redhat.com/articles/3311301

CVE-2017-5715 = Variant 2 of the vulnerability. You can find codes in the next page

https://www.kb.cert.org/vuls/id/584653

 

[-Fran-]

The context of the discussion was Servers. Are those tests done in Server environments?

https://cloudblogs.microsoft.com/microsoftsecure/2018/01/09/understanding-the-performance-impact-of-spectre-and-meltdown-mitigations-on-windows-systems/

Even if what they claim is true, which their are reports to the contrary with just Meltdown patches on linux systems(1 system call mapped into 7). Variant 2 patching causes a further overhead for Intel systems on linux systems. AMD claims that Variant 2 hasn't been shown to affect AMD systems, so they are not patching for it in the microcode. So, we are back to AMD not being affected again.

Variant 1 and Variant 3 mitigations have minimal performance impact, while Variant 2 remediation, including OS and microcode, has a performance impact.

https://www.amd.com/en/corporate/speculative-execution

Differences in AMD architecture mean there is a near zero risk of exploitation of this variant. Vulnerability to Variant 2 has not been demonstrated on AMD processors to date.

Several reports in the Internet reporting supposed Meltdown-patches performance are in reality measuring Spectre-patch impact. E.g. Amazon slowdown is due to Spectre patches, not Meltdown

https://www.realworldtech.com/forum/?threadid=173959&curpostid=173981

And now we have Microsoft confirming that the real problem is Spectre. This is understandable because the 'fix' for Variant 2 of the vulnerability consists on disabling branch prediction, which affects performance.

I know AMD PR claims that their CPUs are safe... but AMD engineers are working with kernel developers to patch the CPUs for Variant 2

https://lists.opensuse.org/opensuse-security-announce/2018-01/msg00004.html
https://www.suse.com/es-es/support/update/announcement/2018/suse-su-20180011-1/
https://bugs.gentoo.org/643476
https://access.redhat.com/articles/3311301

CVE-2017-5715 = Variant 2 of the vulnerability. You can find codes in the next page

https://www.kb.cert.org/vuls/id/584653

And do you have any measurements to clearly state that, for AMD, it will be a performance decrease?

I haven't seen any AMD server benchmarks, for better or for worse, because almost no one has them (yet?). So, it's close to a moot point to even say that AMD will be impacted EQUALLY to Intel based on current data.

Do I need to make the bold word bigger?

 

[goldstone77]

The context of the discussion was Servers. Are those tests done in Server environments?

https://cloudblogs.microsoft.com/microsoftsecure/2018/01/09/understanding-the-performance-impact-of-spectre-and-meltdown-mitigations-on-windows-systems/

Even if what they claim is true, which their are reports to the contrary with just Meltdown patches on linux systems(1 system call mapped into 7). Variant 2 patching causes a further overhead for Intel systems on linux systems. AMD claims that Variant 2 hasn't been shown to affect AMD systems, so they are not patching for it in the microcode. So, we are back to AMD not being affected again.

Variant 1 and Variant 3 mitigations have minimal performance impact, while Variant 2 remediation, including OS and microcode, has a performance impact.

https://www.amd.com/en/corporate/speculative-execution

Differences in AMD architecture mean there is a near zero risk of exploitation of this variant. Vulnerability to Variant 2 has not been demonstrated on AMD processors to date.

Several reports in the Internet reporting supposed Meltdown-patches performance are in reality measuring Spectre-patch impact. E.g. Amazon slowdown is due to Spectre patches, not Meltdown

https://www.realworldtech.com/forum/?threadid=173959&curpostid=173981

And now we have Microsoft confirming that the real problem is Spectre. This is understandable because the 'fix' for Variant 2 of the vulnerability consists on disabling branch prediction, which affects performance.

I know AMD PR claims that their CPUs are safe... but AMD engineers are working with kernel developers to patch the CPUs for Variant 2

https://lists.opensuse.org/opensuse-security-announce/2018-01/msg00004.html
https://www.suse.com/es-es/support/update/announcement/2018/suse-su-20180011-1/
https://bugs.gentoo.org/643476
https://access.redhat.com/articles/3311301

CVE-2017-5715 = Variant 2 of the vulnerability. You can find codes in the next page

https://www.kb.cert.org/vuls/id/584653

Architectural Defaults
By default, each of the 3 tunables that apply to an architecture will be enabled automatically at boot time, based upon the architecture detected.

Intel Defaults:

pti 1 ibrs 1 ibpb 1 -> fix variant#1 #2 #3
pti 1 ibrs 0 ibpb 0 -> fix variant#1 #3 (for older Intel systems with no microcode update available)

AMD Defaults:
Due to the differences in underlying hardware implementation, AMD X86 systems are not vulnerable to variant #3. The correct default values will be set on AMD hardware based on dynamic checks during the boot sequence.

pti 0 ibrs 0 ibpb 2 -> fix variant #1 #2 if the microcode update is applied
pti 0 ibrs 2 ibpb 1 -> fix variant #1 #2 on older processors that can disable indirect branch prediction without microcode updates

On linux the AMD and Intel implementation is clearly different! These are not windows servers! These have shown impact from day one on Intel servers! Note older Intel system, before skylake, are vulnerable to variant 2 with no microcode fix available! And like Yuka mentions. You lack any real world benchmarks of any impact on AMD systems, but their is a plethora of data available for the impact on Intel systems including the ones provided by Intel.
http://www.tomshardware.com/forum/id-3609004/cpu-security-vulnerabilities-information/page-5.html#20585128

Edit: Also:

Only Variant 2, one of the two vulnerabilities that make up Spectre, requires a CPU microcode update. Variant 2 is also the vulnerability that AMD has said it is most likely not vulnerable to, thus the company has not issued any updates.

The only required BIOS updates are to address Variant 2 for Intel CPUs. If your Intel machine is from a system OEM, look for the updates to come from that manufacturer, most of which are linked here. DIY builders are, as usual, left waiting for motherboard OEMs to release updates

http://www.tomshardware.com/forum/id-3609004/cpu-security-vulnerabilities-information/page-5.html#20585128

 

[8350rocks]

I have several contacts in financial datacenter IT arena, and it is looking bad for them. One of them works for a large banking chain in the U.S. and has supposed that they will need approximately 25% more hardware to accommodate normal workloads after the patch based on current internal projections. Willing to guess what he is going to recommend they buy to supplement existing hardware? I will give you 3 guesses, but I am betting you will only need one.

WOW. That is really terrible consequences. I wonder if all companies incurring in such high costs will consider suing the hardware/OS makers. After all some of those companies are much bigger and powerful than intel (not to mention AMD).

Guess 1 :AMD ?

They are considering action, but that is early stages at this point. They have to maintain business, and so they need an immediate fix.

Guess was correct.

 

[8350rocks]

The context of the discussion was Servers. Are those tests done in Server environments?

https://cloudblogs.microsoft.com/microsoftsecure/2018/01/09/understanding-the-performance-impact-of-spectre-and-meltdown-mitigations-on-windows-systems/

Even if what they claim is true, which their are reports to the contrary with just Meltdown patches on linux systems(1 system call mapped into 7). Variant 2 patching causes a further overhead for Intel systems on linux systems. AMD claims that Variant 2 hasn't been shown to affect AMD systems, so they are not patching for it in the microcode. So, we are back to AMD not being affected again.

Variant 1 and Variant 3 mitigations have minimal performance impact, while Variant 2 remediation, including OS and microcode, has a performance impact.

https://www.amd.com/en/corporate/speculative-execution

Differences in AMD architecture mean there is a near zero risk of exploitation of this variant. Vulnerability to Variant 2 has not been demonstrated on AMD processors to date.

Several reports in the Internet reporting supposed Meltdown-patches performance are in reality measuring Spectre-patch impact. E.g. Amazon slowdown is due to Spectre patches, not Meltdown

https://www.realworldtech.com/forum/?threadid=173959&curpostid=173981

And now we have Microsoft confirming that the real problem is Spectre. This is understandable because the 'fix' for Variant 2 of the vulnerability consists on disabling branch prediction, which affects performance.

I know AMD PR claims that their CPUs are safe... but AMD engineers are working with kernel developers to patch the CPUs for Variant 2

https://lists.opensuse.org/opensuse-security-announce/2018-01/msg00004.html
https://www.suse.com/es-es/support/update/announcement/2018/suse-su-20180011-1/
https://bugs.gentoo.org/643476
https://access.redhat.com/articles/3311301

CVE-2017-5715 = Variant 2 of the vulnerability. You can find codes in the next page

https://www.kb.cert.org/vuls/id/584653

MS is just pushing Intel FUD, likely because of the backend deal on surface and other products. MS has a vested interest in seeing Intel *not* take a huge PR hit.

Linux is fine without patches crippling AMD...funny that.

 

[aldaia]

Impact on ESI production servers (EVE on line) after updates
https://twitter.com/CCP_SnowedIn/status/948980181577875456

 

[juanrga]

And do you have any measurements to clearly state that, for AMD, it will be a performance decrease?

I haven't seen any AMD server benchmarks, for better or for worse, because almost no one has them (yet?). So, it's close to a moot point to even say that AMD will be impacted EQUALLY to Intel based on current data.

Do I need to make the bold word bigger?

I don't need measurements to know that patches will affect AMD performance as well. As stated multiple times one of the fixes for Spectre consists on disabling predictive branches for risky code, and disabling branches has a performance impact. Here you have a server benchmark showing performance hit on EPYC after patched for Spectre

You can see that Spectre only on EPYC produces a bigger impact than Meltdown+Spectre on Xeon. Other server benches shows Xeon chips are more affected than EPYC

And other benches show no impact on any.

So this repetitive pretension that Intel is doomed but AMD is safe has to END.

 

[-Fran-]

And do you have any measurements to clearly state that, for AMD, it will be a performance decrease?

I haven't seen any AMD server benchmarks, for better or for worse, because almost no one has them (yet?). So, it's close to a moot point to even say that AMD will be impacted EQUALLY to Intel based on current data.

Do I need to make the bold word bigger?

I don't need measurements to know that patches will affect AMD performance as well. As stated multiple times one of the fixes for Spectre consists on disabling predictive branches for risky code, and disabling branches has a performance impact. Here you have a server benchmark showing performance hit on EPYC after patched for Spectre

Spoiler

You can see that Spectre only on EPYC produces a bigger impact than Meltdown+Spectre on Xeon. Other server benches shows Xeon chips are more affected than EPYC

Spoiler

And other benches show no impact on any.

Spoiler

So this repetitive pretension that Intel is doomed but AMD is safe has to END.

No, what needs to end is you making everyone believe that Intel is not affected by it and everything is OK. You just provided justification that the patches indeed do not affect AMD and Intel equally, yet you've been stating over and over and over that watching Intel's performance take a deep dive is absolutely comparable to what AMD might suffer, which is obviously not true.

And keep in mind these are day-1 patches. These measurements are not "final" either.

 

[juanrga]

MS is just pushing Intel FUD, likely because of the backend deal on surface and other products. MS has a vested interest in seeing Intel *not* take a huge PR hit.

Linux is fine without patches crippling AMD...funny that.

AMD has finally admitted it is affected by and stock is dropping. Benches of patched linux crippling AMD were given by another poster. I copy one of them

 

[goldstone77]

Actually, until there is proof to the contrary Intel is affected more by these patches. How many servers are exclusively using Intel products? Meltdown takes one system call and maps it into 7. That alone is more overhead than AMD since AMD isn't affected by Meltdown. Also, reports now show there is a lot of system instability after the patches, and servers are have to reboot much more often.
I showed the 2 tests form phoronix.com greatest differences in performance between Intel and AMD on patch 4.14 here
http://www.tomshardware.com/forum/id-3609004/cpu-security-vulnerabilities-information/page-5.html#20585746
This clearly shows intel has a much bigger hit to performance out of all the tests vs. AMDs.

The Meltdown & Spectre thread is littered with various services reporting ~30% or greater CPU utilization after the patches on Intel products. We have yet to see any reports from any services of severe CPU utilization from AMD systems, but Intel we have several.

http://www.tomshardware.com/forum/id-3609004/cpu-security-vulnerabilities-information/page-5.html#20585463

http://www.tomshardware.com/forum/id-3609004/cpu-security-vulnerabilities-information/page-4.html#20576594

http://www.tomshardware.com/forum/id-3609004/cpu-security-vulnerabilities-information/page-4.html#20576623
Until we have even one definitive report you are just speculating against synthetic benchmarks, which you leave out the one showing Intel is having the most significant performance loss. I find this rather convenient since you are arguing in favor of Intel. The fact that we don't even have one service reporting about AMD performance issues, and multiple reports of Intel's performance issues that constitutes a greater problem for Intel. And this would make their enterprise customers at the very least consider other options like AMD or ARM.