text stringlengths 11 441 |
|---|
[2033.98 --> 2037.82] Basically, the summary of all of this is that... |
[2038.48 --> 2040.20] And we've done a fair bit of investigation. |
[2040.48 --> 2042.10] We're working on our own overclocking guide. |
[2042.10 --> 2045.44] But when the high-precision event timer is off... |
[2045.44 --> 2049.44] If your machine goes to sleep... |
[2049.44 --> 2051.80] Then wakes up... |
[2051.80 --> 2057.00] Your benchmarks will appear to have higher scores. |
[2057.00 --> 2057.86] Okay? |
[2057.86 --> 2062.24] So, this is actually because the system timer is running slow. |
[2062.76 --> 2070.80] So, Windows' clock, and anything that uses timing is actually slowed by a fair margin. |
[2071.24 --> 2075.14] So, benchmarks, which measure compute time, will score higher. |
[2075.14 --> 2077.92] So, I believe these notes are from Anthony. |
[2078.04 --> 2078.60] Hold on one second. |
[2078.90 --> 2079.66] Hey, Anthony! |
[2080.42 --> 2082.94] Those Cinebench scores, those are your scores, right? |
[2083.66 --> 2084.44] Okay, cool. |
[2084.60 --> 2084.98] Thanks, man. |
[2085.52 --> 2085.76] Okay. |
[2086.14 --> 2091.98] So, we got $17.50 in Cinebench pre-sleep. |
[2092.48 --> 2093.54] Put the machine to sleep. |
[2094.02 --> 2094.86] Woke it back up. |
[2095.12 --> 2096.30] Over $2,000. |
[2096.30 --> 2103.22] So, there seems to be a fair bit of confusion out there right now around Ryzen performance. |
[2103.52 --> 2111.18] Because if people are reporting benchmark scores based on some of the tips and tricks out there |
[2111.18 --> 2115.18] that involve turning off HPET and going to sleep and waking, |
[2115.76 --> 2120.64] those benchmarks are not actually representative of real performance. |
[2120.64 --> 2124.44] To be clear, Ryzen does perform very well. |
[2124.72 --> 2129.64] But what this seems to indicate to us, and actually, hold on, I'm just going to... |
[2130.32 --> 2130.98] Hey, Anthony! |
[2131.78 --> 2132.06] Yep. |
[2132.90 --> 2140.26] Our scores on the crosshair did not change after we disabled HPET, |
[2140.34 --> 2142.02] as long as we didn't go to sleep, right? |
[2143.12 --> 2143.60] Yeah. |
[2143.84 --> 2148.64] So, the one you did go to the first day, that off the time, it did, and then it did. |
[2148.80 --> 2149.22] Got it. |
[2149.30 --> 2149.58] Okay. |
[2149.58 --> 2156.82] So, our review was done with HPET on, because we didn't receive the reviewer guidance |
[2156.82 --> 2159.58] that HPET might be bugged on the crosshair in time. |
[2159.58 --> 2161.72] So, on would be correct, right? |
[2161.86 --> 2162.74] Correct scores? |
[2162.88 --> 2170.34] On means that our scores, while maybe not quite optimal, are at least... |
[2170.34 --> 2171.64] Not affected by the timer bug. |
[2171.64 --> 2176.08] Not affected by the bug that makes them appear artificially inflated. |
[2176.32 --> 2176.52] Okay. |
[2176.96 --> 2178.00] So, there you go. |
[2178.00 --> 2179.00] Yeah. |
[2179.00 --> 2180.00] Yeah. |
[2180.00 --> 2189.28] So, the window sleep bug makes AMD Ryzen overclock higher and gives better benchmark results. |
[2189.28 --> 2196.26] So, it requires further investigation, which is why our OC guide is coming out probably next week or the week after, |
[2196.56 --> 2198.34] and not this week. |
[2198.34 --> 2203.34] So, like, as a relative layman, like, what would cause something like that? |
[2203.40 --> 2204.48] Do we have any idea? |
[2204.48 --> 2204.54] Yeah. |
[2205.54 --> 2217.78] Basically, the scuttlebutt is that AMD could have delayed the launch of Ryzen and squashed some of these types of bugs. |
[2217.78 --> 2229.74] These bugs don't appear to cause risk of user data loss or anything devastating like that, |
[2229.74 --> 2234.78] but that there are some definite early adopter... |
[2234.78 --> 2236.22] Quirks. |
[2236.22 --> 2243.96] Quirks and oddities that could have been fairly easily dealt with if they had put it off a little bit longer. |
[2244.96 --> 2247.00] So, that is sort of the... |
[2247.00 --> 2248.00] That's our takeaway. |
[2248.46 --> 2250.32] And, you know, why... |
[2250.32 --> 2257.70] I know the question's coming, because it's a question that I ask and we debated internally, and it's a fair question. |
[2257.70 --> 2262.30] Why not report on this in the Ryzen launch review? |
[2262.82 --> 2267.60] And the reason for that is because we can say with a very high degree of confidence |
[2267.60 --> 2273.72] that smart people like AMD and Asus are going to get their heads together. |
[2274.12 --> 2279.64] They're going to get feedback from early adopter users who, again, it is my personal belief, |
[2279.92 --> 2285.66] know or should know what they're getting themselves into buying a brand new CPU on a brand new chipset, |
[2285.66 --> 2291.10] running a brand new memory controller, brand new PCIe controller, brand new brick and everything. |
[2291.70 --> 2298.22] It is my belief that AMD and Asus will solve little weird bugs. |
[2298.90 --> 2299.92] In short order. |
[2300.46 --> 2307.40] That review, covering the performance of the Ryzen 1800X, has to sit there for the next decade, |
[2307.70 --> 2309.04] if all goes according to plan. |
[2309.42 --> 2312.96] And I'm not going to spend, you know, 20% of my video, |
[2312.96 --> 2317.06] the precious little time that I have to talk to you guys, |
[2317.44 --> 2321.66] talking about some weird crap we ran into, that by the time you buy one, |
[2321.72 --> 2327.36] by the time it gets put in a box and gets transported on a truck to your house |
[2327.36 --> 2329.60] and you plug it in, is going to be irrelevant. |
[2330.32 --> 2333.92] So that's why we tend not to overdo it on that stuff. |
[2333.92 --> 2337.40] And the concern is obviously there with, you know, |
[2337.74 --> 2343.36] what if this is causing people to make purchasing decisions based on, you know, |
[2343.48 --> 2345.78] inflated benchmarks, numbers, et cetera, et cetera, et cetera. |
[2346.56 --> 2348.22] Which our numbers are not inflated. |
[2348.30 --> 2348.46] Yep. |
[2348.96 --> 2350.00] Which our numbers are not inflated. |
[2350.02 --> 2356.88] And the other concern is causing people to avoid purchasing something that is legitimately very good |
[2356.88 --> 2361.20] because of a very small issue that will be ironed out in a number of days. |
[2361.30 --> 2361.42] Yeah. |
[2362.60 --> 2362.92] So. |
[2363.10 --> 2364.96] Or weeks or whatever, whatever it ends up being. |
[2365.16 --> 2366.48] That's, that's where we're at. |
[2366.82 --> 2367.08] Okay. |
[2367.18 --> 2367.40] Okay. |
[2367.42 --> 2369.34] Another thing people can stop messaging me about. |
[2369.46 --> 2370.38] Paladon 123. |
[2371.12 --> 2372.46] Canadians don't like this. |
[2372.48 --> 2372.76] Okay. |
[2373.22 --> 2374.58] What time is it in Canada? |
[2374.58 --> 2376.38] We have, we have four and a half time zones. |
[2376.50 --> 2376.66] Okay. |
[2377.98 --> 2379.84] It is many times in Canada. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.