3 ms·

We are talking about server chips? Can you point to some comparisons?

kllrnohj 7y ago
> We are talking about server chips? What gave you that idea? This thread was clearly about someone building a local PC. There's no reason to assume that'd be a server chip? But even in the server world go look up any comparison of Epyc Rome vs. Intel Xeon and you'll still struggle to find a reason to go Intel. The Epyc 7742 is in a league of its own at this point.
- spectramax 7y ago
  I’m trying to get a point across for favoritism of AMD beyond objective measures. This entire post is about Intel security flaw where we just had one discovered for AMD yesterday! But people are not as vile about AMD but become hostile when it comes to Intel. Why? There are some toxic people defending both sides with passion. I’m baffled. If this isn’t “Fanboys”, what is? Regarding comparisons, I’m yet to see a conclusive source that points at AMD > Intel for server workloads - benchmarking these things is complex. Power, efficiency, workload types, etc matter. So it’s not as simple as saying “Hey, AMD won”. I’m disappointed at the hostility for pointing out fanboyism on HN. It’s present whether one denies it or not. Edit: Holyshit. I won’t be responding anymore. This has become an unproductive discussion simply for pointing out the bias in supporting the “underdog”. I have no way to prove this but I stand by my observations and what my eyes see - given Apples and Apples comparison for metric X or security flaw X, people have a deep bias for supporting the “underdog”. I saw this first hand back in 2006 when Intel was the “underdog”. This is a fact and no amount of massaging on HN or the Internet forums will change that. I’m again disappointed in the HN community’s response.
  - bayindirh 7y ago
    It sounds like I'm trying to debunk you, but I just want to answer your questions. I work as an HPC administrator and I just want to tell a single thing: We were unable to buy first generation EPYC processors because almost all of the production run is bought by Google, Amazon and Dropbox. They weren't offered by many vendors as general availability servers because it wasn't possible. It's a bit late here so, I cannot do much research but I'll leave this Phoronix article [0] about RocksDB performance. Phoronix Test Suite sounds a bit cheesy but, as far as I can see, it's becoming a reputable benchmark suite. I was troubleshooting a server from a big manufacturer and, its official live CD has this thing installed to see how system performs. Also our fellow HPC centers says that EPYC is not vaporware in terms of performance. [0]: https://www.phoronix.com/scan.php?page=article&item=intel-amd-rocksdb&num=2 https://www.phoronix.com/scan.php?page=article&item=intel-am...
  - bmelton 7y ago
    > I’m disappointed at the hostility for pointing out fanboyism on HN Because it very much comes across as though you've got an axe to grind. OP was "maybe sort of" going to build an AMD box, for which there are a lot of good reasons irrespective of the bug on topic, and this set him over the top. His post doesn't really exude fanboy behavior, and your comment - while possibly appropriate for some other post - isn't really appropriate to OP.
    - spectramax 7y ago
      I don't know if I have an axe to grind - I've said time and again my response had nothing to do with the OP. It was a general observation. I should have actually put it up at the root level instead of responding to OP. Did you see my messages about "general observation"?
  - kllrnohj 7y ago
    > where we just had one discovered for AMD yesterday AMD's was a new side channel that doesn't appear to be that different from existing side channels like flush+reload. Some pros (and cons) over the established ones, but you're not attacking anything with those either. It's how you ex-filtrate data after a successful attack. This, by comparison, is a new attack vulnerability. They are different severity, and I'd expect here of all places to recognize & acknowledge that. I'm baffled as to why you seem to be expecting something different? > Regarding comparisons, I’m yet to see a conclusive source that points at AMD > Intel for server workloads - benchmarking these things is complex. Power, efficiency, workload types, etc matter. So it’s not as simple as saying “Hey, AMD won”. Did you look? Yes benchmarking these is complex, but when one CPU wins the overwhelming majority of every comparison on every metric it reeks of fanboyism to then claim it's "not conclusive." There's no shortage of server CPU benchmarks done in this space. If you have a specific comparison that's missing then say it. Otherwise you're just accusing people of the fanboy behavior that you're exhibiting. But here: https://www.anandtech.com/show/14694/amd-rome-epyc-2nd-gen/15 https://www.anandtech.com/show/14694/amd-rome-epyc-2nd-gen/1... "For those with little time: at the high end with socketed x86 CPUs, AMD offers you up to 50 to 100% higher performance while offering a 40% lower price. Unless you go for the low end server CPUs, there is no contest: AMD offers much better performance for a much lower price than Intel, with more memory channels and over 2x the number of PCIe lanes. These are also PCIe 4.0 lanes. What if you want more than 2 TB of RAM in your dual socket server? The discount in favor of AMD just became 50%. So has AMD done the unthinkable? Beaten Intel by such a large margin that there is no contest? For now, based on our preliminary testing, that is the case. The launch of AMD's second generation EPYC processors is nothing short of historic, beating the competition by a large margin in almost every metric: performance, performance per watt and performance per dollar." Like every review of Rome is like this. AMD released an incredible product. Credit where credit is due is not fanboyism.
  - samatman 7y ago
    There are certain things that get consistently hammered on HN. Referring to people as "fanboys" is one of them. There's a reason for this: it's an insult, for starters. Furthermore, it leads to unproductive discussions. More heat than light. There's almost always a way to express yourself that doesn't involve being dismissive and insulting. You may not realize that's what you're doing by saying "fanboy". But it is what you're doing. If you'd pulled a highfalutin' word like "partisanship" you would probably find a productive conversation. Just avoid fanboy, circlejerk, and other obvious flamebait, and you'll be fine.
bmelton 7y ago
I mean, TFA is about an exploit exclusive to Intel SGX processors, which of course includes E3 Xeons.
- saagarjha 7y ago
  No, it's not. The paper chooses SGX because it's easy to write a PoC against, but there's nothing in it limiting it to SGX. (It's a general technique against any kind of memory load in a privileged context.)

> We are talking about server chips? What gave you that idea? This thread was clearly about someone building a local PC. There's no reason to assume that'd be a server chip? But even in the server world go look up any comparison of Epyc Rome vs. Intel Xeon and you'll still struggle to find a reason to go Intel. The Epyc 7742 is in a league of its own at this point.

spectramax 7y ago
I’m trying to get a point across for favoritism of AMD beyond objective measures. This entire post is about Intel security flaw where we just had one discovered for AMD yesterday! But people are not as vile about AMD but become hostile when it comes to Intel. Why? There are some toxic people defending both sides with passion. I’m baffled. If this isn’t “Fanboys”, what is? Regarding comparisons, I’m yet to see a conclusive source that points at AMD > Intel for server workloads - benchmarking these things is complex. Power, efficiency, workload types, etc matter. So it’s not as simple as saying “Hey, AMD won”. I’m disappointed at the hostility for pointing out fanboyism on HN. It’s present whether one denies it or not. Edit: Holyshit. I won’t be responding anymore. This has become an unproductive discussion simply for pointing out the bias in supporting the “underdog”. I have no way to prove this but I stand by my observations and what my eyes see - given Apples and Apples comparison for metric X or security flaw X, people have a deep bias for supporting the “underdog”. I saw this first hand back in 2006 when Intel was the “underdog”. This is a fact and no amount of massaging on HN or the Internet forums will change that. I’m again disappointed in the HN community’s response.
- bayindirh 7y ago
  It sounds like I'm trying to debunk you, but I just want to answer your questions. I work as an HPC administrator and I just want to tell a single thing: We were unable to buy first generation EPYC processors because almost all of the production run is bought by Google, Amazon and Dropbox. They weren't offered by many vendors as general availability servers because it wasn't possible. It's a bit late here so, I cannot do much research but I'll leave this Phoronix article [0] about RocksDB performance. Phoronix Test Suite sounds a bit cheesy but, as far as I can see, it's becoming a reputable benchmark suite. I was troubleshooting a server from a big manufacturer and, its official live CD has this thing installed to see how system performs. Also our fellow HPC centers says that EPYC is not vaporware in terms of performance. [0]: https://www.phoronix.com/scan.php?page=article&item=intel-amd-rocksdb&num=2 https://www.phoronix.com/scan.php?page=article&item=intel-am...
- bmelton 7y ago
  > I’m disappointed at the hostility for pointing out fanboyism on HN Because it very much comes across as though you've got an axe to grind. OP was "maybe sort of" going to build an AMD box, for which there are a lot of good reasons irrespective of the bug on topic, and this set him over the top. His post doesn't really exude fanboy behavior, and your comment - while possibly appropriate for some other post - isn't really appropriate to OP.
  - spectramax 7y ago
    I don't know if I have an axe to grind - I've said time and again my response had nothing to do with the OP. It was a general observation. I should have actually put it up at the root level instead of responding to OP. Did you see my messages about "general observation"?
- kllrnohj 7y ago
  > where we just had one discovered for AMD yesterday AMD's was a new side channel that doesn't appear to be that different from existing side channels like flush+reload. Some pros (and cons) over the established ones, but you're not attacking anything with those either. It's how you ex-filtrate data after a successful attack. This, by comparison, is a new attack vulnerability. They are different severity, and I'd expect here of all places to recognize & acknowledge that. I'm baffled as to why you seem to be expecting something different? > Regarding comparisons, I’m yet to see a conclusive source that points at AMD > Intel for server workloads - benchmarking these things is complex. Power, efficiency, workload types, etc matter. So it’s not as simple as saying “Hey, AMD won”. Did you look? Yes benchmarking these is complex, but when one CPU wins the overwhelming majority of every comparison on every metric it reeks of fanboyism to then claim it's "not conclusive." There's no shortage of server CPU benchmarks done in this space. If you have a specific comparison that's missing then say it. Otherwise you're just accusing people of the fanboy behavior that you're exhibiting. But here: https://www.anandtech.com/show/14694/amd-rome-epyc-2nd-gen/15 https://www.anandtech.com/show/14694/amd-rome-epyc-2nd-gen/1... "For those with little time: at the high end with socketed x86 CPUs, AMD offers you up to 50 to 100% higher performance while offering a 40% lower price. Unless you go for the low end server CPUs, there is no contest: AMD offers much better performance for a much lower price than Intel, with more memory channels and over 2x the number of PCIe lanes. These are also PCIe 4.0 lanes. What if you want more than 2 TB of RAM in your dual socket server? The discount in favor of AMD just became 50%. So has AMD done the unthinkable? Beaten Intel by such a large margin that there is no contest? For now, based on our preliminary testing, that is the case. The launch of AMD's second generation EPYC processors is nothing short of historic, beating the competition by a large margin in almost every metric: performance, performance per watt and performance per dollar." Like every review of Rome is like this. AMD released an incredible product. Credit where credit is due is not fanboyism.
- samatman 7y ago
  There are certain things that get consistently hammered on HN. Referring to people as "fanboys" is one of them. There's a reason for this: it's an insult, for starters. Furthermore, it leads to unproductive discussions. More heat than light. There's almost always a way to express yourself that doesn't involve being dismissive and insulting. You may not realize that's what you're doing by saying "fanboy". But it is what you're doing. If you'd pulled a highfalutin' word like "partisanship" you would probably find a productive conversation. Just avoid fanboy, circlejerk, and other obvious flamebait, and you'll be fine.

bmelton 7y ago

I mean, TFA is about an exploit exclusive to Intel SGX processors, which of course includes E3 Xeons.

saagarjha 7y ago
No, it's not. The paper chooses SGX because it's easy to write a PoC against, but there's nothing in it limiting it to SGX. (It's a general technique against any kind of memory load in a privileged context.)