2019 iMac - i5 vs i9? Cooling and Throttling?

[macduke]

I finally got the fans to ramp up to a moderate level while doing "work". I still have that 4K Warp Stabilizer running in the background in Premiere Pro. I hooked up a 4K external display and started playing Baby Driver in 4K with the volume down so I could listen for the fans. I opened up the App Store, Apple News, a bunch of tabs in Safari, RadarScope with a 30 frame super-res loop updating in the background, Messages, Photos downloading 22,000 photos and videos from iCloud, Reminders, Notes, a Flywheel local web server (uses VirtualBox), Coda 2, CodeKit, 1Password, plugged in my drives to run a Time Machine backup, and the fans finally kicked up when I loaded my Lightroom Catalog of 35,000 RAW photos and it started syncing smart previews down from Creative Cloud. It's not loud at all and has even gotten a little quieter while writing this.

 

[igilphoto]

So i9 with the 580 is a bad idea ?
I feel stuck,I want a quite iMac but don’t need the Vega .I actually want to buy the 580 just because it’s same price as the 575 in my country,but hesitate to order because of the heat and noise that may occur.

 

[fokmik]

if you want a quiet imac, you should go with the base i5 3.0Ghz and 570X, thats the only option for this generation...like the last gen the 90W+ make this thing hot...or wait for the next gen with an interior redesign, if Apple is willing to give us

 

[amoergosum]

"Searching for Throttles on the 2019 8 Core i9 iMac"

"The 2017 iMac apparently throttled, alot. Since the design had not changed, everyone assumed this would throttle as well under some level of load. And so I recorded this but could not see any throttling of any significance at all. I pegged the cpus for 20 minutes and nothing. I ran Cinebench and nothing. Not sure what else to try, but the problem that existed on previous versions appear to be resolved."

 

[igilphoto]

"Searching for Throttles on the 2019 8 Core i9 iMac"

"The 2017 iMac apparently throttled, alot. Since the design had not changed, everyone assumed this would throttle as well under some level of load. And so I recorded this but could not see any throttling of any significance at all. I pegged the cpus for 20 minutes and nothing. I ran Cinebench and nothing. Not sure what else to try, but the problem that existed on previous versions appear to be resolved."

Hi,is it with the Vega or 580?

 

[Bboble]

"Searching for Throttles on the 2019 8 Core i9 iMac"

"The 2017 iMac apparently throttled, alot. Since the design had not changed, everyone assumed this would throttle as well under some level of load. And so I recorded this but could not see any throttling of any significance at all. I pegged the cpus for 20 minutes and nothing. I ran Cinebench and nothing. Not sure what else to try, but the problem that existed on previous versions appear to be resolved."

Seeing hardcore throttling. Intel rates the 9900k at 4.7GHz when all cores are pegged. The video shows a wiggly line (indicative of throttling) and, a frequency of ~3.8GHz. Furthermore, the Cinebench score should be about 1000 points higher on a stock 9900k. At full tilt, the 9900k also uses about twice the wattage.

 

[Colonel Blimp]

Hi,is it with the Vega or 580?

Vega 48. The same YouTuber has an unboxing video here.

 

[Chancha]

Seeing hardcore throttling. Intel rates the 9900k at 4.7GHz when all cores are pegged. The video shows a wiggly line (indicative of throttling) and, a frequency of ~3.8GHz. Furthermore, the Cinebench score should be about 1000 points higher on a stock 9900k. At full tilt, the 9900k also uses about twice the wattage.

What is curious to me is that, many owners are reporting fan noise level are down from the 2017 model, if that's true then it means Apple decided to not fire the fan as fast as previous gen, while at the same time feels ok for the 9900k to not achieve its potential performance.

 

[fokmik]

who are those many owners that they had 2017 and now bought the 2019 and says that?

 

[ondert]

As I expected, they probably locked down the cpu to 95w limit, and that OpenCL score from Vega 48... (facepalm was here) it’s a minor upgrade from RP580. I don’t know about RP580X yet.

 

[Rockadile]

Seeing hardcore throttling. Intel rates the 9900k at 4.7GHz when all cores are pegged. The video shows a wiggly line (indicative of throttling) and, a frequency of ~3.8GHz. Furthermore, the Cinebench score should be about 1000 points higher on a stock 9900k. At full tilt, the 9900k also uses about twice the wattage.

95W limited 9900K will do all-core 3.6GHz so either Apple raised the limit a little higher or the PSU is not good enough to supply enough wattage.
[doublepost=1553850957][/doublepost]

What is curious to me is that, many owners are reporting fan noise level are down from the 2017 model, if that's true then it means Apple decided to not fire the fan as fast as previous gen, while at the same time feels ok for the 9900k to not achieve its potential performance.

Looks like the same gentle fan curve as iMac Pro. The temp is pegged at 92-93C. Spikes to 100C on Cinebench.

 

[Chancha]

who are those many owners that they had 2017 and now bought the 2019 and says that?

Not literally, just 2nd hand conjecture. From the individual reviews and even hands on videos, the fan noise does seem lesser than what I have experienced on my 2017, which I can easily make it roar without trying.

Either way I haven't seen a single person posting a fan RPM history graph. Need to wait for people who actually knows how to conduct a test.

 

[Bboble]

95W limited 9900K will do all-core 3.6GHz so either Apple raised the limit a little higher or the PSU is not good enough to supply enough wattage.

The 95 watt TDP for the 9900k is meaningless. At full tilt, it pulls about twice this. It's also rated to run at 4.7GHz full tilt. 3.6GHz is just the base frequency. Apple should be partnering with Intel and getting their best bin, and undervolting the processor.
[doublepost=1553851644][/doublepost]

Not literally, just 2nd hand conjecture. From the individual reviews and even hands on videos, the fan noise does seem lesser than what I have experienced on my 2017, which I can easily make it roar without trying.

Either way I haven't seen a single person posting a fan RPM history graph. Need to wait for people who actually knows how to conduct a test.

Curious on this, as well. They may have changed the fan design to be less audible, yet flow more CFM...they may have lowered the wattage consumed to 85 watts, etc... Curious to put the speculation to an end.

 

[igilphoto]

So basically in order to maintain a quieter machine ,with less noise they limited the i9 and limited the power ?Did I understood correctly?

That’s kind of sad if so (instead of designing better cooling ) but in character for apple I think ,performance is less important than the illusion of silence and looks.

I’m ok with the i9 not doing all that it supposed to do in other computeres ,as long it still significantly better than the i5 option and not going to be too hot/malfunction just because apple want it to be silent .
I’m still waiting for reviews and currently looking at the the i9 with 580 and ssd.Will buy the i5 if people will report that the i9 is too hot/loud.I have no need for the Vega for my usage .


What do you guys think ?

 

[Rockadile]

The 95 watt TDP for the 9900k is meaningless. At full tilt, it pulls about twice this. It's also rated to run at 4.7GHz full tilt. 3.6GHz is just the base frequency. Apple should be partnering with Intel and getting their best bin, and undervolting the processor.
[doublepost=1553851644][/doublepost]

Curious on this, as well. They may have changed the fan design to be less audible, yet flow more CFM...they may have lowered the wattage consumed to 85 watts, etc... Curious to put the speculation to an end.

It's meaningless for a regular 9900K but this is clearly acting similar to one with it's power limited to 95W. I don't see an unrestricted 9900K going all-core 4.7GHz in an AIO.

 

[Trusteft]

So basically in order to maintain a quieter machine ,with less noise they limited the i9 and limited the power ?Did I understood correctly?

No. They are just guessing.

 

[patearrings]

.

 

[Rockadile]

So basically in order to maintain a quieter machine ,with less noise they limited the i9 and limited the power ?Did I understood correctly?

That’s kind of sad if so (instead of designing better cooling ) but in character for apple I think ,performance is less important than the illusion of silence and looks.

I’m ok with the i9 not doing all that it supposed to do in other computeres ,as long it still significantly better than the i5 option and not going to be too hot/malfunction just because apple want it to be silent .
I’m still waiting for reviews and currently looking at the the i9 with 580 and ssd.Will buy the i5 if people will report that the i9 is too hot/loud.I have no need for the Vega for my usage .


What do you guys think ?

Seeing as the i9 CPU speed doesn't dip really low and back up constantly, it will still be worth it.
Apple has decided to keep the CPU hot when it's working hard for low noise. These Intel chips are pretty durable so I'm more worried about nearby components failing when above 90C then the CPU.

 

[Colonel Blimp]

Seeing as the i9 CPU speed doesn't dip really low and back up constantly, it will still be worth it.
Apple has decided to keep the CPU hot when it's working hard for low noise. These Intel chips are pretty durable so I'm more worried about nearby components failing when above 90C then the CPU.

What I’m curious about is whether the Vega 48 will throttle when the CPU is running hot.

For example, if I recall correctly, the Radeon M295X in the 2014 iMac was reported to throttle down under heavy load, but that problem went away in the 2015 iMacs. My Radeon M395X never throttles down, even when it’s running at 100°C.

If the Radeon Pro Vega 48 can maintain its performance without throttling, with the i9 cores maintaining a steady ~3.9 GHz, and with the fans remaining as quiet as those in my 2015 iMac, I’ll be quite satisfied.

 

[hydr]

So basically in order to maintain a quieter machine ,with less noise they limited the i9 and limited the power ?Did I understood correctly?

That’s kind of sad if so (instead of designing better cooling ) but in character for apple I think ,performance is less important than the illusion of silence and looks.

I’m ok with the i9 not doing all that it supposed to do in other computeres ,as long it still significantly better than the i5 option and not going to be too hot/malfunction just because apple want it to be silent .
I’m still waiting for reviews and currently looking at the the i9 with 580 and ssd.Will buy the i5 if people will report that the i9 is too hot/loud.I have no need for the Vega for my usage .


What do you guys think ?

Then how do you explain the 33000 geekbench score people are getting? That is a major upgrade over the 2017 update.

 

[ondert]

Then how do you explain the 33000 geekbench score people are getting? That is a major upgrade over the 2017 update.

Geekbench calculates your cpu's capacity. It doesn't put the cpu under load. So, if the cpu can not be cooled well, that Geekbench score is simply useless. I stop considering Geekbench scores after 2018 mbp started to throttle heavily.

 

[joema2]

...Then how do you explain the 33000 geekbench score people are getting? That is a major upgrade over the 2017 update.

Geekbench calculates your cpu's capacity. It doesn't put the cpu under load. So, if the cpu can not be cooled well, that Geekbench score is simply useless...

It's not just Geekbench. In the above-referenced video, he also ran CineBench R20 many times back to back, and he only heard the fans on 2nd and subsequent runs, and only for 15 sec.

When I run CineBench R20 on my i7 2017 iMac 27, I hear the fans after 30 sec, and they are loud at 60 sec -- on the *initial* run. The fans stay audible consistently from 30 sec to the benchmark completion at 2 min 15 sec. His R20 results were 3887, my i7 2017 iMac did 2286, my 10-core iMac Pro did 4396. The iMP stays quiet no matter how many R20 runs.

Unlike past versions, CineBench R20 is designed to test whether the machine's cooling solution is sufficient for longer-running tasks to deliver the full potential of the CPU: https://www.maxon.net/en/products/cinebench-r20-overview/

It appears the 2019 8-core i9 iMac 27 has better thermal and fan acoustic performance under high load than the 2017 i7 iMac. Just how much better will be revealed as more in-depth tests are done.

 

[quagmire]

Seeing hardcore throttling. Intel rates the 9900k at 4.7GHz when all cores are pegged. The video shows a wiggly line (indicative of throttling) and, a frequency of ~3.8GHz. Furthermore, the Cinebench score should be about 1000 points higher on a stock 9900k. At full tilt, the 9900k also uses about twice the wattage.

Yes an unrestricted 9900K with proper cooling can sustain all core of 4.7 GHz. When Anandtech limited it to 95W, all core would only sustain 3.6 GHz. If one was expecting an unrestricted 9900K in the iMac’s enclosure, think those expectations were set wrong.

 

[kaintxu]

As I expected, they probably locked down the cpu to 95w limit, and that OpenCL score from Vega 48... (facepalm was here) it’s a minor upgrade from RP580. I don’t know about RP580X yet.

How is the vega 48 only a minor upgrade from the 580? 580 has a open CL of 110k ves the almost 150k of the vega?
[doublepost=1553865493][/doublepost]

Yes an unrestricted 9900K with proper cooling can sustain all core of 4.7 GHz. When Anandtech limited it to 95W, all core would only sustain 3.6 GHz. If one was expecting an unrestricted 9900K in the iMac’s enclosure, think those expectations were set wrong.

How does a restricted 9900k compared to a regular 9600k

 

[jonatious]

Yes an unrestricted 9900K with proper cooling can sustain all core of 4.7 GHz. When Anandtech limited it to 95W, all core would only sustain 3.6 GHz. If one was expecting an unrestricted 9900K in the iMac’s enclosure, think those expectations were set wrong.

If Apple is limiting to 3.6GHz at 8 cores, then why the hell are they advertising Boost upto 5GHz?

/e Understood the difference