Intel's new Microcode patch is HERE! Impact Testing Performance...

JayzTwoCents
8 Aug 202418:01

Summary

TLDRThis video script discusses the release of the latest microcode for MSI motherboards, specifically for Intel CPUs, ahead of the expected schedule. The script explores the performance implications and behavioral changes of the new microcode, focusing on Intel's 14th Gen 900K CPU. It includes tests on an MSI MPG Z790 Carbon 2 Wi-Fi motherboard, comparing performance before and after the microcode update. The results show minor fluctuations in performance, with some tests showing slight uplifts or drops, but no significant overall impact. The script also addresses concerns about CPU stability and the potential need for further updates, emphasizing the importance of long-term testing to ensure the microcode effectively addresses instability issues.

Takeaways

  • 📅 MSI motherboards were the first to release the latest microcode updates for Intel CPUs ahead of schedule, expected in mid-August.
  • 🔍 The microcode update, specifically 0x1129, is intended to address stability issues in Intel's 13th and 14th generation CPUs, but it will not fix CPUs already experiencing crashes.
  • 💻 Falcon Northwest, a high-end custom gaming PC builder, is highlighted for their focus on performance and quality, offering systems with a three-year warranty and shipping coverage.
  • 🛠️ The script discusses the necessity of updating both the BIOS and the Management Engine (ME) to properly apply the microcode to the CPU.
  • 📊 Performance tests show mixed results post-microcode update, with some tests showing minor uplifts, others showing slight drops, and some remaining nearly identical.
  • ⚠️ There is skepticism about whether the microcode update is a permanent fix or a temporary Band-Aid, indicating the need for long-term testing.
  • 📉 In certain tests like Time Spy Extreme, there was a noticeable performance drop, suggesting potential changes in CPU behavior post-update.
  • 🎮 Gaming performance tests showed minimal impact from the microcode update, with games like Shadow of the Tomb Raider showing no significant change.
  • 🌡️ Temperature and power consumption remained largely the same after the update, contrary to expectations of potential reductions due to voltage adjustments.
  • 🔋 Voltage spikes during single-core workloads were observed to be lower post-update, which could indicate a more stable and safer operating condition for the CPU.
  • 🤔 The final assessment is that while there may be minor performance hits, stability is preferred, and the true effectiveness of the microcode update will only be known over time.

Q & A

  • What is the main topic discussed in the video script?

    -The main topic discussed in the video script is the release and testing of the latest microcode updates for Intel CPUs, specifically focusing on the performance and stability implications for 14th and 13th generation Intel Core CPUs.

  • Which motherboard brand was first to release the microcode updates according to the script?

    -MSI was the first motherboard brand to release the microcode updates for Intel CPUs.

  • What is the purpose of the microcode updates for Intel CPUs?

    -The purpose of the microcode updates is to address crashing and instability issues experienced by some Intel CPUs, particularly those from the 13th and 14th generation with 65 watts and up.

  • What is the microcode version that the script mentions as the latest?

    -The latest microcode version mentioned in the script is 0x1129.

  • How does the script describe the performance impact of the microcode updates on Intel CPUs?

    -The script suggests that there is no significant performance hit from the microcode updates, with most tests showing minimal changes and some even showing slight performance improvements in certain scenarios.

  • What is the significance of the BIOS update mentioned in the script?

    -The BIOS update is significant because it includes the microcode update, which is necessary for addressing the stability issues with Intel CPUs. It also involves changes to the Intel Extreme and Intel Performance profiles.

  • What additional step is required besides updating the BIOS for the microcode to take effect?

    -Besides updating the BIOS, the Management Engine (ME) installer must also be run to ensure that the microcode is applied to the CPU.

  • What is the sponsored brand mentioned in the script, and what do they offer?

    -The sponsored brand mentioned in the script is Falcon Northwest, which offers custom gaming PCs with a focus on high-end gaming experiences, custom cases, and a three-year warranty policy.

  • What is the script's stance on the long-term effectiveness of the microcode updates in resolving CPU stability issues?

    -The script expresses skepticism about the long-term effectiveness of the microcode updates, suggesting that it will take time to determine if the updates truly resolve the stability issues or are just a temporary fix.

  • What advice does the script give to users experiencing CPU crashes related to the microcode issues?

    -The script advises users experiencing CPU crashes related to microcode issues to perform an RMA (Return Merchandise Authorization) process, as their CPU is likely already damaged.

  • How does the script describe the voltage behavior before and after the microcode update?

    -The script describes that before the microcode update, there were voltage spikes close to 1.6 volts, particularly in single-core workloads. After the update, these spikes were reduced and did not exceed 1.51 volts, indicating a change in voltage management.

Outlines

00:00

🔍 Early Release of MSI Microcode Update

The script discusses the unexpected early release of the latest microcode for MSI motherboards compatible with Intel CPUs. The update, initially anticipated for mid-August, was released ahead of schedule with MSI being the first to roll it out. The primary focus is on the potential impact of the update on performance and behavior changes, particularly when tested on Intel's 14th Gen 900K CPU. The script also introduces a sponsored segment for Falcon Northwest, a high-end custom gaming PC builder, emphasizing their 30-year history, custom cases, and rigorous testing to ensure optimal performance. The company offers a comprehensive warranty and shipping coverage, inviting viewers to explore their offerings through a provided link.

05:00

📉 Performance Impact of New Microcode on Intel CPUs

This paragraph delves into the performance implications of the new microcode 0x1129, which is an update to address stability issues in Intel's 13th and 14th generation CPUs. The narrator clarifies that this update will not remedy instability already present in affected CPUs, suggesting it's an Intel issue that requires a more permanent solution. The script details the testing process on an MSI MPG z790 Carbon 2 Wi-Fi motherboard with BIOS updates and the ME engine installer, highlighting the necessity of both for the microcode to take effect. Performance tests, including Cinebench R23 and R24, Geekbench, and various other benchmarks, reveal minor fluctuations in performance, with some tests showing slight improvements and others minor declines, indicating no significant overall impact.

10:04

🎮 Gaming Performance and Microcode Update Effects

The paragraph examines the effects of the microcode update on gaming performance, using titles like Shadow of the Tomb Raider and Cyberpunk 2077 to illustrate any potential drops in frames per second (FPS). While Shadow of the Tomb Raider showed no significant change due to the CPU and GPU combination reaching the game's engine cap, Cyberpunk 2077 demonstrated a slight decrease in average and minimum FPS. The narrator also discusses the importance of single-core performance and voltage sensitivity, which can affect stability and potentially cause damage to the CPU. The paragraph concludes with a detailed analysis of core clock speeds, package temperatures, and voltage spikes before and after the microcode update, showing a reduction in voltage spikes post-update.

15:06

🔧 Stability Concerns and the Future of Intel Microcode

The final paragraph addresses the community's concerns regarding the stability and trustworthiness of Intel's microcode updates. The narrator suggests that while the performance hit from the update is minimal (around 2% on average), the real test will be the long-term stability of the CPUs. They mention the possibility of further microcode revisions, as evidenced by the quick succession of releases from MSI, and emphasize the need for caution and patience to see if the updates truly resolve the underlying issues. The script concludes by acknowledging the challenges faced by users with unstable CPUs, recommending RMA as a last resort and expressing hope that the new microcode will prove effective over time.

Mindmap

Keywords

💡Microcode

Microcode is a layer of machine-level code used in CPUs that helps translate instructions into operations within the processor. In the context of the video, the latest microcode update for Intel CPUs is discussed, which is intended to address stability issues and performance concerns. The script mentions that MSI motherboards were the first to receive these updates, and the video aims to assess the impact of these microcode revisions on system performance.

💡MSI Motherboards

MSI is a well-known manufacturer of computer hardware, including motherboards. The script highlights that MSI was the first to release the new microcode updates for Intel CPUs. The motherboards are central to the testing process described in the video, as they provide the platform on which the CPU performance is evaluated after the microcode update.

💡Intel CPUs

Intel CPUs refer to the central processing units manufactured by Intel Corporation. They are the primary focus of the video, as the script discusses a microcode update intended to improve the stability and performance of these processors. The video specifically tests the Intel Core i9-12900K processor to observe any changes in behavior or performance post-update.

💡Performance Stability

Performance stability in the context of CPUs refers to the processor's ability to maintain consistent operation without crashes or instability. The video discusses the expectation that the new microcode will improve stability, as previous versions of Intel's 13th and 14th generation CPUs were reported to have stability issues.

💡Behavioral Changes

Behavioral changes refer to the modifications in the way a system operates after an update or modification. In the script, the term is used to describe the potential shifts in CPU performance characteristics following the microcode update, such as changes in clock speeds, voltage requirements, and thermal performance.

💡BIOS Update

A BIOS (Basic Input/Output System) update is a software revision that can improve system functionality or compatibility. The script mentions that along with the microcode update, a BIOS update was necessary to ensure the new microcode is properly applied to the CPU. The video tests different BIOS versions to observe the impact on CPU performance.

💡ME Engine Installer

The ME (Management Engine) Engine Installer is a software tool used to update the Management Engine firmware on Intel-based systems. The script specifies that updating the ME Engine is a separate process from the BIOS update and is necessary for the microcode to take effect on the CPU.

💡Cinebench R23

Cinebench R23 is a benchmarking software that assesses the performance of CPUs by running a series of tests. The script uses Cinebench R23 to evaluate the performance of the Intel CPU before and after the microcode update, noting any differences in scores as an indicator of the update's impact.

💡Voltage Spikes

Voltage spikes refer to sudden increases in electrical voltage within a system. The video discusses the observation of voltage spikes in the CPU during certain workloads, which can potentially lead to instability or damage. The new microcode is expected to mitigate these voltage spikes and thus improve CPU stability.

💡Package Power

Package power is the total power consumption of the CPU package, including all cores and uncore components. The script mentions observing changes in package power after the microcode update, suggesting a smoother power delivery that may contribute to improved stability and thermal performance.

💡Gaming Performance

Gaming performance is a measure of how well a system can run games, typically assessed by frame rates and graphical fidelity. The video includes gaming benchmarks to see if the microcode update affects the CPU's ability to handle CPU-intensive titles like 'Cyberpunk 2077', which relies heavily on CPU processing power.

Highlights

Latest microcode for MSI motherboards and Intel CPUs released ahead of schedule.

MSI was the first to release the microcode update, which was initially expected in mid-August.

Performance and behavioral changes discussed regarding Intel's 14th Gen 900K CPU after microcode update.

Microcode 0x1129 is newer than the previously leaked 0x125, addressing crashing and instability issues in 13th and 14th Gen CPUs.

If a 13th or 14th Gen CPU is already unstable, the new microcode will not fix it; it's considered an Intel issue.

For stable CPUs, the new microcode aims to maintain stability, though long-term testing is needed to confirm effectiveness.

MSI MPG z790 Carbon 2 Wi-Fi motherboard was used for testing with two different BIOS versions.

The new BIOS update includes Intel Extreme and Intel Performance profiles, which were not in the initial release.

Both BIOS and ME engine updates are required for the microcode to take effect on the CPU.

Cinebench R23 showed no significant performance change after the microcode update.

Cinebench R24 multi-threading showed a noticeable drop in performance after the update.

Geekbench scores showed a minor drop, within the margin of error, after the microcode update.

Handbrake render time showed a negligible increase, possibly within the margin of error.

Blender benchmark showed a slight performance improvement after the update.

Time Spy showed a significant drop in CPU test score after the microcode update.

Gaming performance in Shadow of the Tomb Raider remained nearly identical after the update.

Cyberpunk 2077 showed a slight drop in FPS, indicating a potential impact of the microcode update on CPU-intensive titles.

Voltage spikes during single-core workloads were reduced after the microcode update.

Package power showed a smoother fluctuation pattern after the update, suggesting more stable voltage delivery.

The impact of the microcode update on performance is minor, with an average hit of about 2%.

The community's trust in Intel's microcode update is uncertain, and long-term stability remains to be seen.

Transcripts

play00:00

okay we got the latest micro code at

play00:01

least for the MSI motherboards that

play00:02

we're going to check out regarding Intel

play00:04

CPUs uh this is kind of ahead of

play00:06

schedule actually we were expecting

play00:07

middle of August for these micro codes

play00:09

to start dropping MSI was first and we

play00:12

tested it we're going to talk about what

play00:13

kind of performance you may or may not

play00:15

lose and what sorts of Behavioral

play00:17

changes there are with Intel's at least

play00:19

14 900k which is what we tested on first

play00:22

time time time to pay the

play00:24

bills for those looking for a high-end

play00:26

custom gaming experience look no further

play00:28

than Falcon Northwest Falcon Northwest

play00:30

has been building PCS made for gamers

play00:32

for over 30 years with a focus on a true

play00:34

high-end gaming experience custom cases

play00:36

available only through Falcon Northwest

play00:38

feature state-of-the-art testing and

play00:39

design to ensure that every component is

play00:41

performing at their best through thermal

play00:43

imaging and rigorous lab testing

play00:44

designed and overseen by the Falcon

play00:46

Northwest founder himself with a

play00:47

complete lineup of systems ranging from

play00:49

small to large every Falcon Northwest

play00:51

system includes a three-year warranty

play00:52

policy and a year of 2-way overnight

play00:54

shipping coverage providing the ultimate

play00:56

peace of mind to see all that Falcon

play00:57

Northwest has to offer follow the

play00:59

sponsored link in the description

play01:01

below all right so I'm talking about

play01:03

micro code 0x1 129 it's actually later

play01:07

or newer than 0 x125 which is the code

play01:11

or the micro code that was sort of

play01:12

leaked that intel was currently testing

play01:14

for the crashing and instability

play01:16

problems regarding your 14th 13th gen C

play01:18

KC CPUs anything 65 watts and up which

play01:21

is just about all of them at this point

play01:23

um spoiler alert we've already talked

play01:25

about this if your 13th and 14th gen CPU

play01:27

is already experiencing and exhibiting

play01:29

in stability Behavior it's done it's

play01:32

toast this code is not going to fix it

play01:34

that is an Intel problem that they are

play01:35

going to have to figure out how the hell

play01:37

they're going to deal with that but this

play01:39

is for those of you that have had or

play01:41

have stable CPUs like mine included at

play01:43

home has not had any crashing issues and

play01:44

I've been running it at 320 watt for a

play01:46

long time this problem does not

play01:48

necessarily plague every single CPU I

play01:51

mean the micro code stuff there's

play01:53

variant in there

play01:54

so this is for those of you that do not

play01:56

have crashing Intel CPUs and you want to

play01:58

keep them that way this is thetically

play02:00

the way that's supposed to work although

play02:01

I think we're all a little leery of

play02:03

whether or not it's really going to fix

play02:04

the problem and it's not just a Band-Aid

play02:06

so long-term testing is really going to

play02:08

show over time if these are going to fix

play02:10

the problems so we ran the 14900 K on

play02:14

our MSI MPG z790 Carbon 2 Wi-Fi

play02:17

motherboard not an Asus board so the two

play02:19

bios for the carbon 2 that we used here

play02:21

one of them goes back to May which is

play02:24

e7d 89 im.

play02:27

a42 um which is the first bios MS I

play02:30

launched was which specifically put the

play02:32

Intel extreme and Intel performance

play02:34

profiles in the Bios feed to select

play02:36

prior to that none of those profiles

play02:37

were there at all they defaulted to the

play02:39

4,096 Watts the 511 amps the whole deal

play02:42

right the unlimited power which was what

play02:45

was part of the problem with all of

play02:46

these CPU deaths so we ran through our

play02:48

CPU test and then we loaded the latest

play02:52

bios which actually dropped today this

play02:54

is a very fast video we're working on

play02:56

this but today which is actually a few

play02:58

days later than the previous bios which

play03:00

dropped with the 0 x125 which is the

play03:03

first revision micro code fix from Intel

play03:05

so MSI has been rolling these out

play03:07

actually pretty quickly so this bios the

play03:09

second bios was the e7d 89 im. A51 so on

play03:13

their website as of

play03:15

888224 now that could change and there

play03:17

could still be some revisions to that

play03:19

micro code but we wanted to see what

play03:20

kind of performance hit increase changes

play03:24

Behavior whatever that's going to happen

play03:26

by changing uh the micro code now the

play03:28

micro code and the me engine also had a

play03:30

separate installer that we had to run so

play03:32

it wasn't just the BIOS update you need

play03:34

to look at the package uh from your bios

play03:37

manufacturer or your motherboard

play03:38

manufacturer and see what their patching

play03:40

notes are because if you were to just

play03:42

install the BIOS but not run the me

play03:44

engine or the me installer you would not

play03:46

have gotten the micro code on your CPU

play03:48

either so you have to run those together

play03:50

Asus motherboards tends to do it all

play03:52

together as a package it'll do a restart

play03:53

and then do the Emy automatically so

play03:55

depending on your motherboard you may or

play03:56

may not have to do that so check with

play03:58

your motherboards installation

play04:00

instructions let's go and get right into

play04:02

the results here because this is where

play04:04

things get kind of interesting because

play04:07

you'll notice if we look in the chart

play04:08

here and I left all our CPUs on there

play04:10

we'll start with um we'll start with

play04:12

cinebench R23 all core so you see at the

play04:16

bottom of the chart it says 14,900 K 0

play04:18

x29 that's the latest bios and the 0 x23

play04:22

that's what the micro code was for the

play04:24

BIOS that came out with the performance

play04:26

and extreme profiles it's also important

play04:27

to note this is on the extreme 250 53

play04:30

wat profile which is actually what Intel

play04:32

says the performance is supposed to be

play04:35

but as you can see no significant change

play04:38

visible uh to the 14900 K in that test

play04:40

if you take a look at cinebench R23 the

play04:43

single thread score um again a slight

play04:46

uplift 2336 on single core to a 2345

play04:50

single core again margin of error you

play04:52

can run this 10 10 this run this test 10

play04:54

times and get 10 different results there

play04:56

if we look at r24 which is an actually a

play04:59

harder

play05:00

test to run than R23 it does it's

play05:02

different types of instructions that are

play05:03

actually happening during the test it's

play05:05

not just one that's hammering the cores

play05:07

uh we went from a

play05:09

2136 down to a 2124 I would actually say

play05:12

in the multi-threading test in r24 this

play05:15

is actually more of a significant

play05:16

measurable drop um a 2136 to a 2124 is

play05:20

noticeable you'll notice these are much

play05:22

lower score numbers than like R23 so a

play05:25

22 point or a 12o drop is actually

play05:28

somewhat noticeable um in terms of the

play05:30

test now if we go to a single core run

play05:33

on r24 it's exactly the same 136 to 136

play05:37

you'll notice kind of a theme Here the

play05:39

single core stuff doesn't really seem to

play05:40

be affected too much although single

play05:42

core workloads is where a lot of the

play05:44

voltage would jump too high as high as

play05:46

1.6 on the old micro code um where you

play05:49

actually would see the voltage problem

play05:51

more than just all core if you enable an

play05:53

allcore workload most of the cores due

play05:55

to vrou will actually drop their vid

play05:58

drop their vid out of the the danger

play06:00

zone but it's actually the single core

play06:01

workloads that have a a chance to burst

play06:04

the voltage high enough to potentially

play06:05

cause damage um at least I had a whole

play06:07

discussion with Falcon Northwest about

play06:09

that which was actually quite surprising

play06:10

I always just assumed the all core

play06:12

workload would be the biggest stress to

play06:13

the CPU but no if we're talking about

play06:16

voltage sensitivity it's actually single

play06:18

and two core workloads that can cause

play06:20

that burst and we'll look at the chart

play06:21

later where you'll actually see some of

play06:23

that on the pre um 0x1 129 all right so

play06:26

looking at geekbench here a little bit

play06:28

of a drop so from 2,169 to a

play06:31

20,33 it's important to note this is all

play06:33

the exact same system for for this these

play06:36

tests this is not like we threw it on a

play06:38

different rig and saw how it performed

play06:40

this is only the BIOS and management

play06:42

engine change for the micro code now

play06:44

geekbench uh single core you can see we

play06:45

go from a 3209 to a

play06:48

3197 um fairly margin of error there not

play06:51

the significant performance hit at least

play06:53

that we've seen so far that a lot of

play06:54

people were expecting potentially happen

play06:56

I think what people were thinking was

play06:57

going to happen is that there was going

play06:59

to maybe be

play07:00

with the new micro code a voltage SL

play07:03

frequency drop across the board on these

play07:06

on the cores and so far that's not what

play07:08

we're seeing here I don't think I don't

play07:09

think we most of us ever expected that

play07:11

to be the case but I know that that's

play07:13

what people have been sort of like

play07:14

throwing out into the into the void

play07:16

Sphere for people to either just I don't

play07:19

know believe or whatnot but if we look

play07:20

at handbrake you can see we gained one

play07:23

second of render time so handbreak is a

play07:25

lower is better type of score it's how

play07:26

many seconds it took to transcode our 4K

play07:29

video down to 1080p um we're not using

play07:32

Quicks sync or any of that stuff this is

play07:34

just raw CPU power and you can see we

play07:36

went from 59 seconds to a 60 seconds

play07:38

again complete margin of error that

play07:40

could be a rounding error with the way

play07:41

that does a fraction of a second it

play07:42

could have been more than 59.5 which at

play07:45

that point would just round up to 60

play07:46

seconds now we actually picked up some

play07:47

performance in blender so you can see we

play07:50

went from in the classroom render um

play07:52

which is a r Trace rendering we went

play07:53

from 11055 to a

play07:56

11127 a little bit larger than the

play07:58

margin of error we've seen now these are

play07:59

samples samples higher is better um we

play08:02

tend to see that we can run this test 10

play08:04

times and get very very very similar

play08:06

results every single time not this like

play08:09

two or three point variant so there was

play08:10

a little bit of a pickup there if we

play08:13

take a look at uh junk shop however we

play08:15

lost a little bit of performance there

play08:16

1589 to

play08:18

15628 again potentially margin of error

play08:21

but it's still noticeable and and we

play08:23

have to mention that now monster is

play08:24

nearly identical at

play08:26

22928 versus

play08:28

22955 so nearly identical performance

play08:31

there now time spy was an interesting

play08:32

one because all these tests are

play08:34

different type of instruction that we're

play08:35

asking the CPU to do and different types

play08:37

of instructions will hit the cores and

play08:38

hit the logic with the way the voltage

play08:40

and the frequencies operate in a

play08:42

different way so Times by extreme went

play08:44

from 11,941 down to an

play08:47

11,264 700 points in the CPU test is a

play08:50

significant drop that's actually

play08:52

noticeable sometimes we have to

play08:54

overclock our CPUs to pick up 700 points

play08:57

in CPU tests in time stream when

play08:59

we're doing like the xoc stuff so seeing

play09:01

it drop 700 points is actually a a

play09:04

pretty significant talking point right

play09:06

here that that so far is the biggest

play09:08

change we've seen with the new micro

play09:11

code versus the existing micro code now

play09:13

if we go into gaming however shadow of

play09:15

the Tomb Raider pretty much identical

play09:18

the reason for that is the fact that the

play09:19

14900 K and an RTX 490 can hit the

play09:22

engine cap of 300 Mega 300 frames per

play09:25

second in Shadow of the Tomb Raider so

play09:28

it's 1080p medium no scaling no RTX

play09:30

higher is better we're trying to push as

play09:32

many frames to the CPU as possible so

play09:35

that we can see how the CPU handles that

play09:36

so you can see 301 and 300 300's a new

play09:39

cap that game used to be capped at

play09:41

around 220 over the years they've

play09:42

actually increased that cap now it's

play09:44

around 300 so you can see we're hitting

play09:46

the cap before we're hitting or seeing

play09:48

any sort of degradation in performance

play09:49

by having this new micro code cyberpunk

play09:52

2077 however is a CPU intensive title

play09:56

and we did see a drop although you'd

play09:58

have a hard time noticing it we did see

play10:00

a drop 236 average FPS down to

play10:03

229 and our minimum FPS also dropped

play10:06

from a 190 to a

play10:08

186 now like I said now like I said

play10:12

cyberpunk is a more intensive CPU type

play10:15

of title it's got a lot of AI happening

play10:17

in the game it's got a huge City that's

play10:19

got a lot of NPCs doing things and the

play10:21

CPU is in charge of handling a lot of

play10:23

the way that those uh simulations

play10:25

interact in the game and then obviously

play10:27

the GPU isn't isn't responsible for all

play10:29

of the like post-processing of the

play10:30

scenes to make them beautiful and

play10:32

obviously accelerate the scenes so a

play10:34

actual measurable drop in cyberpunk um

play10:38

2077 in fact it dropped it almost down

play10:41

to the 14600 K number now obviously we

play10:45

have to retest all of our 14th gen and

play10:48

13th gen CPUs prior to the next round of

play10:50

CPU testing because all of these CPUs

play10:52

may see the same potential type of drop

play10:54

most of the problems that we've been

play10:56

experiencing has specifically been on

play10:58

the 900 are the I9 SKS so 13900 and

play11:01

14900 some 13700 and 14 700s have seen

play11:05

instability issues very few of the 14600

play11:08

or 13600 CPUs have seen this type of

play11:10

issue it really comes down to the higher

play11:13

more dense core count CPUs that are

play11:15

having this problem now seeing FPS

play11:16

charts and stuff like that that's only

play11:18

part of the story the other thing we did

play11:19

was I did a 10-minute cinch run R23

play11:22

allcore um to just sort of see if

play11:24

there's any visual differences that

play11:26

happen over time so I charted out over

play11:28

time a few different things so for

play11:29

instance we'll start with the core

play11:30

clocks here um what I'm looking for here

play11:33

is to see if there's any core clock drop

play11:35

was there any drop between pre1 129 or

play11:38

post1 129 and as you can see by looking

play11:40

at this chart the green and the blue are

play11:42

the Pees the orange and the yellow are

play11:43

the EES you can see we're still getting

play11:46

about 5.1 5.15 megahertz uh or gigahertz

play11:50

on all the peores before and after and

play11:53

all of the trends are the same all these

play11:55

low spikes that you see by the way this

play11:56

is when the test restarts so the

play11:59

frequency actually drops during the test

play12:01

restart um and then obviously under load

play12:03

they jump back up because we do have C

play12:05

States enabled in the CPU that's part of

play12:06

the profile stuff to make sure C states

play12:08

are enabled um believe it or not a lot

play12:10

of the motherboards actually were

play12:11

disabling C States so they would just

play12:13

run full speed constantly all the time

play12:15

so that's why you see these sudden drops

play12:16

in the chart now if you look at the

play12:18

start though on the ecor pre0

play12:23

x12900 MHz higher for the first three

play12:27

and a half runs and then it basically

play12:29

matches where we are post 0x1 129 so I

play12:33

thought that that was interesting that

play12:34

we see no difference on the PE cor but

play12:36

we see a little behavioral difference on

play12:37

the ecor now if we move over here to

play12:40

package temperatures they were pretty

play12:42

much identical now what I was hoping to

play12:44

see here was that we'd see some um

play12:46

temperature drop because if we were

play12:47

seeing potential voltage dropping then

play12:50

that would lead to lower temperatures as

play12:53

you can see here with our um my Cooler

play12:55

Master 280 uh millimeter AIO on here we

play12:59

pretty much have identical temperatures

play13:01

across the board so the green line as

play13:02

you can see here this is package temp

play13:04

pre micro code the blue line is post

play13:07

micro code and it's pretty much

play13:09

identical now if we take a look at the

play13:11

voltage on the other hand um this is

play13:14

where things are a little bit

play13:16

interesting because if you take a look

play13:17

at the green line which is the pre1 129

play13:20

patch or micro code patch you can see

play13:23

the voltage spikes so you saw in the

play13:25

frequency test in between the tests the

play13:27

frequency drops however there there's

play13:29

going to be one or two cores and

play13:30

remember there's 28 threads in a 14900 K

play13:34

one or two cores is going to spike to

play13:36

load the test that's just because of the

play13:37

fact that it becomes a like very low

play13:40

thread workload to load the test or get

play13:42

it started so this is when we're going

play13:44

to see these voltage spikes now look at

play13:46

the spikes right especially the ones

play13:48

right here in the middle they get

play13:49

awfully close to 1.6 volts while

play13:53

starting the test and then you can see

play13:55

during the test they shoot down and then

play13:57

as the load comes off and it reloads the

play13:58

test test that single core voltage Spike

play14:01

spikes again real high and it's not that

play14:03

high at the start it's what's

play14:05

interesting about that is it's not that

play14:06

high at the start of the test it starts

play14:08

to go higher in the middle of the test

play14:10

now if you take a look at the blue line

play14:11

which is post micro code update you can

play14:14

see we still have those similar spikes

play14:16

in the same spots but they're not as

play14:17

high they're not exceeding 1.51 volts

play14:21

was the highest that I observed whereas

play14:23

we were getting 1.58 volts prior to it

play14:26

so this is where we're actually seeing

play14:28

the micro code changes is going to be in

play14:30

single core um and maybe low core count

play14:33

type of workloads where those CPUs are

play14:35

staying at a much higher core clock

play14:37

closer to the 6 GHz which is what you're

play14:39

supposed to get on single and two core

play14:41

performance on the 900

play14:43

SKS to get 6 GHz you need voltage to

play14:46

support that so as you can see right

play14:48

here it definitely drops the voltage a

play14:51

little bit not a lot but a little bit

play14:53

after the micro code update so this is

play14:55

where we saw the biggest change was this

play14:58

the the local core count voltage spikes

play15:00

now if we take a look at package power

play15:02

this is also where it's a little bit

play15:03

interesting too the green line as you

play15:06

can see it's steady at 253 and when I

play15:09

say steady I mean it's there's a little

play15:11

bit of a wave in there but it is like

play15:12

you can tell 253 is the cap however the

play15:14

blue line as you can see has a more kind

play15:17

of a wavy fluctuating pattern underneath

play15:20

253 so it actually dropped down to about

play15:24

245 to like 248 average with spikes 253

play15:30

so that's probably linked obviously to

play15:32

voltage voltage is directly correlated

play15:35

to Watts obviously voltage watts and

play15:38

amps all three of those are related

play15:39

that's how electricity works so it makes

play15:41

sense with that little bit of voltage

play15:43

change that we were seeing at least with

play15:45

the low core count voltages that there

play15:47

would be a different way that it hits

play15:49

the Headroom or the the target I should

play15:51

say so it hits that Target a little more

play15:53

smoothly rather than just throwing

play15:55

itself against that 253 watt limit and

play15:58

then just locking itself there so this

play16:01

is the way it looks post micro code now

play16:04

as I said already whether or not you

play16:07

trust Intel's micro code to really be

play16:09

the fix it's probably the best idea to

play16:12

wait and see how this goes over time the

play16:16

like I said MSI was the first brand to

play16:18

come out right now with the micro code I

play16:20

got a million emails this morning when I

play16:22

woke up saying hey the new micro code's

play16:23

out why don't you test it here it is at

play16:26

the end of the day it is not and and I

play16:29

know this sounds like Intel fanboying

play16:31

right now it's not I'm team computer I

play16:33

love computers I don't care if it's

play16:34

Intel or AMD or

play16:37

Nvidia I just I just love computers

play16:39

whatever it is that works for you but

play16:42

it's not the oh my God your CPUs are

play16:44

about to get nerfed type of performance

play16:46

hit at most we saw about 2% of average

play16:48

performance hit no one's going to notice

play16:50

2% if I have to give up 2% or even 5% of

play16:53

performance to have it be stable yeah

play16:55

I'll be upset about that because that

play16:57

would mean I lost some performance that

play16:58

I paid for and was advertised to get but

play17:02

I'd much rather have my CPU be stable we

play17:04

took bigger hits with Spectre and

play17:05

meltdown with those updates which were

play17:07

security updates than we're seeing with

play17:09

this but again I don't think the story

play17:11

is over yet I I I because of the fact

play17:14

that MSI released a0 x125 and then only

play17:16

a few days later a 0x1

play17:18

129 I'm wondering if there's like a0x

play17:21

131 or something Intel is working on now

play17:24

we need to see over time if this fixes

play17:26

the problem like I said if you have a

play17:29

CPU that's not crashing you're probably

play17:30

safe but that doesn't mean it hasn't

play17:32

already degraded somewhat that you that

play17:34

isn't noticeable because your workloads

play17:36

don't hit it in a way that it would

play17:37

present itself that's the trust issue

play17:39

that everyone's having right now with

play17:41

Intel and unfortunately as I said if

play17:43

you're already getting blue screens and

play17:44

stuff and it's related to this your CPU

play17:46

is already toast it's time to RMA that

play17:49

however painful that process is going to

play17:51

be all right guys there you go tell me

play17:52

how you feel about the new micro code

play17:54

hopefully this fixes stuff but we

play17:56

definitely aren't going to know right

play17:57

now it's going to take time thanks for

play17:58

watching as always we'll see you in the

play17:59

next one

Rate This

5.0 / 5 (0 votes)

Etiquetas Relacionadas
Microcode UpdateCPU PerformanceIntel CPUsMSI MotherboardsSystem StabilityGaming PCsFalcon NorthwestCustom GamingBIOS UpdateVoltage Stability
¿Necesitas un resumen en inglés?