|
I was hoping it'd be more in the Intel latency ballpark. Oh well.
|
#
?
Mar 17, 2018 04:46
|
|
|
|
| # ? Apr 25, 2024 20:57 |
|
|
GRINDCORE MEGGIDO posted:I was hoping it'd be more in the Intel latency ballpark. Oh well. Still a really nice improvement, and Ryzen L1/L2/L3 has much better bandwidth which in theory should help with the worse latency.
|
|
#
?
Mar 17, 2018 05:02
|
|
|
Those latency improvements don't affect the IF, right?
|
|
#
?
Mar 17, 2018 15:28
|
|
|
Combat Pretzel posted:Those latency improvements don't affect the IF, right?  Sisoftware accidentally put their review up ahead of time. Edit: looks like the CPUs were tested on stock clocks with turbo enabled and with DDR4 at 2400. Arzachel fucked around with this message at 17:29 on Mar 17, 2018 |
|
#
?
Mar 17, 2018 17:22
|
|
|
Do people compare with 6700K due to similar IPC, or what's the reason?
|
|
#
?
Mar 17, 2018 17:55
|
|
|
Yeah thats a little odd now that you mention it. How about some of the new 6 core Intels to compare with..
|
|
#
?
Mar 17, 2018 18:01
|
|
|
redeyes posted:Yeah thats a little odd now that you mention it. How about some of the new 6 core Intels to compare with.. Because the new 6/8/10 core chips use a mesh topography vs. ring, which can do interesting things to the inter-core latency.
|
|
#
?
Mar 17, 2018 20:45
|
|
|
Coffee Lake hexcore uses a mesh, too? I think it's still on the ringbus. The mesh applies to the -X CPUs.
|
|
#
?
Mar 17, 2018 20:55
|
|
|
Yeah CFL hexcores still use the ring bus
|
|
#
?
Mar 17, 2018 20:56
|
|
|
Most likely they just didn't have any Coffee Lake chips on hand to run the test suite in time.
|
|
#
?
Mar 17, 2018 21:52
|
|
|
https://twitter.com/HardwareUnboxed/status/975613747271266304 Great video except it was a little hilarious to see them talking about how the difference in 1t performance of like 50fps in Overwatch didn’t matter. But at least they showed the full data without going full Adored.
|
|
#
?
Mar 19, 2018 13:32
|
|
|
ufarn posted:https://twitter.com/HardwareUnboxed/status/975613747271266304 I don't know... seems fine to be hitting 244Hz at 1080p Ultra instead of 300Hz.
|
|
#
?
Mar 19, 2018 13:39
|
|
|
AMD clearing out the stock before Zen+ arrives with some Dealz. Still probably wise to check Microcenter first. https://www.anandtech.com/show/12539/amd-and-partners-slash-prices-on-select-ryzen-ryzen-threadripper-cpus
|
|
#
?
Mar 19, 2018 21:47
|
|
|
Now if we could see some RAM price c... ahahaha the best I can hope for is some $15 class action lawsuit rebate in 2022 for "buying one or more RAM modules during the price collusion period during 2017-2018"
|
|
#
?
Mar 20, 2018 05:10
|
|
|
Alpha Mayo posted:Now if we could see some RAM price c... ahahaha the best I can hope for is some $15 class action lawsuit rebate in 2022 for "buying one or more RAM modules during the price collusion period during 2017-2018" Samsung: "That's a nice DRAM fab we got there. It would be a shame if something happened to it."
|
|
#
?
Mar 20, 2018 06:33
|
|
|
You joke, but Samsung just had a power cut at one of their NAND fabs. 30 minutes of cut power = 60,000 dead wafers, or 11% of the entire month's quota.
|
|
#
?
Mar 20, 2018 09:35
|
|
|
SwissArmyDruid posted:You joke, but Samsung just had a power cut at one of their NAND fabs. 30 minutes of cut power = 60,000 dead wafers, or 11% of the entire month's quota. I know that too well since I was the one who first posted that news here. The timing was impeccable too: NAND prices were already in a slight decline this quarter and the damage suspiciously convenient and reversible. A deliberate 3.5% loss in global monthly supply is a loving grade AAA investment strategy if it meant the cartel gets to easily raise prices for more than that for rest of the year. Palladium fucked around with this message at 10:31 on Mar 20, 2018 |
|
#
?
Mar 20, 2018 10:25
|
|
|
cartel? I think you mean "trio that has opted for co-existence as the best way to maximize profitability"
|
|
#
?
Mar 20, 2018 10:36
|
|
|
Alpha Mayo posted:cartel? I think you mean "trio that has opted for co-existence as the best way to maximize profitability" What do you think a cartel is?
|
|
#
?
Mar 20, 2018 11:34
|
|
|
whoosh?
|
|
#
?
Mar 20, 2018 11:37
|
|
|
Wouldn’t be the first time.
|
|
#
?
Mar 20, 2018 11:42
|
|
|
AMD has confirmed+detailed the CTS Labs exploits and says they'll be patched in a few weeks. Notably, it turns out the PSP compromise actually does bypass the signature authentication. The chipset exploit may bypass signing also, they don't say if an attacker with kernel control could disable driver authentication in a running Windows instance but I don't see why it would be categorically impossible. Kernel mode is kernel mode, the system is owned at that point. Paul MaudDib fucked around with this message at 21:47 on Mar 20, 2018 |
|
#
?
Mar 20, 2018 21:33
|
|
|
AMD's Obituary!
|
|
#
?
Mar 20, 2018 21:36
|
|
|
Poor AMD. Just when we thought they finally got their poo poo together on the CPU side of things.
|
|
#
?
Mar 20, 2018 21:56
|
|
|
<picks up a box, walks out the door with it> "Hey look guys, I discovered a security flaw, this box isn't welded to the floor! Am I a security expert now?" As has been said before, while there *is* a legitimate security flaw there, you have bigger problems if you let someone get to that point. SwissArmyDruid fucked around with this message at 22:06 on Mar 20, 2018 |
|
#
?
Mar 20, 2018 21:59
|
|
|
Right now you can use the 15% off coupon on eBay to get the Asrock X399M for $290, oh yeah and a $10 rebate too  $700 is still a little too rich for my blood, especially with a newer version coming later this year, but it's tempting. Paul MaudDib fucked around with this message at 22:48 on Mar 20, 2018 |
|
#
?
Mar 20, 2018 22:46
|
|
|
I was really hoping for the exploits to be bullshit. Not for any sort of "lol nice meltdown intel xdddd" fanboyism, but because I was honestly hoping CTS Labs was just some lovely stock manipulation scam. That is a preferable alternative to some rowdy shits getting away with 24 hour disclosure on firmware level privilege escalation vulnerabilities.
|
|
#
?
Mar 20, 2018 23:45
|
|
|
To be fair, it's not like "well, there _are_ some bits wrong, but..." exactly exonerates them for their behavior or justifies the entire stock-manipulating venture. They were pretty shady in GN's followup questions, too.
|
|
#
?
Mar 20, 2018 23:54
|
|
|
To me it always looked like the CEO and CTO are guys not far removed from college who were somewhat competent at hacking and they discovered something in ASMedia's chipset, and the CTO's brother who took finance courses told them that they can make money from this and "I'll take it from here", and brought marketing and public relations to make things look slick and phony-professional. And in doing so, they turned over the operation to the guy who isn't technical and doesn't know how security firms actually operate. It's even more obvious when you look at their letter and their Anand interview and it's the guy with the hedge fund background who fields all inquiries about why they didn't handle it similarly to other security firms. It's a classic Woz/Jobs kind of thing, where the Wozniak nerdy guy has a eureka moment and his smooth marketing-savvy buddy envisions startup gold.
|
|
#
?
Mar 21, 2018 00:09
|
|
|
Except the Steves revolutionized an industry, and these guys are just smug cunts looking for the quick buck. fakeedit: so actually that fits Jobs pretty well
|
|
#
?
Mar 21, 2018 00:12
|
|
|
https://www.youtube.com/watch?v=DxQ0PCFsitI
|
|
#
?
Mar 21, 2018 00:12
|
|
|
So an attacker with kernel level access to the system can do kernel level things with it? I still don't see the exploit. You can also flash a corrupted Intel Management Engine into the BIOS and brick an Intel computer or permenantly change configuration settings most people aren't even aware of, is that also an "exploit"?
|
|
#
?
Mar 21, 2018 01:21
|
|
|
Alpha Mayo posted:So an attacker with kernel level access to the system can do kernel level things with it? I still don't see the exploit. You can also flash a corrupted Intel Management Engine into the BIOS and brick an Intel computer or permenantly change configuration settings most people aren't even aware of, is that also an "exploit"? Mmyep. I would not be surprised if chunks of their PoC is copy/pasted from the Zen BIOS and Kernel Developer's Guide. Obituary indeed.
|
|
#
?
Mar 21, 2018 02:10
|
|
|
How does the process for getting a potentially seg faulting ryzen replaced work? Do I need to send my CPU in first or?
|
|
#
?
Mar 21, 2018 11:34
|
|
|
Alpha Mayo posted:So an attacker with kernel level access to the system can do kernel level things with it? I still don't see the exploit. You can also flash a corrupted Intel Management Engine into the BIOS and brick an Intel computer or permenantly change configuration settings most people aren't even aware of, is that also an "exploit"? The firmware validation bypass is a legitimate problem whether or not there was disclosure fuckery. The fuckery just made the whole situation a little more dangerous, so it shouldn't be glossed over.
|
|
#
?
Mar 21, 2018 14:57
|
|
|
The kind of changes in firmware these vulns allow shouldn't be permitted short of cracking the hardware open and attempting to re-flash by soldering on to the serial pins and even that should be protected by some mechanism. The people here saying "root user can make root changes" are being dumb as poo poo and there is a big difference between having full control of the OS vs arbitrary modification of its underlying hardware including security-critical components like the TPM. Conflating those is lazy, garbage analysis. This isn't nearly as bad as Meltdown allowing drive-by javascript to dump the entire contents of memory from a trivial exploitability standpoint but this type of vulnerability is a goldmine for state-level actors if they can get physical access for a few minutes because it means a permanent backdoor that will likely never be discovered by the target.
|
|
#
?
Mar 21, 2018 18:24
|
|
|
It's a real vulnerability, and a serious vulnerability. It should have been responsibly disclosed. It isn't in the same league as Meltdown and Spectre, it probably didn't deserve a web page and press blitz.
|
|
#
?
Mar 21, 2018 19:13
|
|
|
lol okay 'serious vulnerability'? If an attacker has root access to your system, they can flash your BIOS with garbage, they can install ransomware, they can install a rootkit. They OWN your system at that point. which loving attacker is going to do firmware fuckery when they have literally EVERY OTHER MEANS to do whatever the gently caress they want meanwhile Spectre/Meltdown allow reading kernel memory from loving javascript running under a VM. Yeah totally comparable. might as well call "physically taking a computer" a 'serious vulnerability' too. I mean in that case they can just take the CPU out of the socket and run off with it! try booting with no CPU!
|
|
#
?
Mar 21, 2018 19:27
|
|
|
Except with this you're taking a machine you've already exploited and are coding more backdoors into the ARM chip that runs even when disabled and isn't user-facing because it's supposed to be a discreet security feature, and thus you're not going to have some kind of antivirus scanner that can tell you if your PSP was hacked. So those backdoors are coded in there, will survive reboots, will survive OS reformats. The best way to describe it that I can think of: Meltdown is horrible for everyone, including the average guy logging into his bank at home. Whereas the Ryzen vulnerabilities are the kind of thing the guy has no need to be concerned about, but the bank should be very concerned about. Same thing with cloud providers and any other data-driven business where there's enough employees that you couldn't tell if someone in the hall works here everyday or is an NSA employee installing an exploit and giving the system's administrator a gag order under threat of treason and aiding terrorism.
|
|
#
?
Mar 21, 2018 19:36
|
|
|
|
| # ? Apr 25, 2024 20:57 |
|
|
~~FOR IMMEDIATE RELEASE~~ An attacker with physical access to systems running AMD based processors, including Ryzen and Epyc, may be vulnerable to permanent damage. In out proof of concept, an attacker was able to remove the CPU, and scratch up the motherboard with a car key they had on hand. The system was rendered permanently unbootable. ALL AMD CHIPS, RELEASED FROM 1982 AND ON, ARE EFFECTED! AMD has yet to provide resolution to this matter but are investigating the attack vector. -company totally not set up overnight by stock market manipulators hoping computer algorithms read "negative AMD news story" and drop the price
|
|
#
?
Mar 21, 2018 19:36
|
|