But WHERE on the Intel spec sheet lists this? I don’t see their designs state a 4.7Ghz minimum clock speed or a TDP higher than 95w. Therefore, Apple’s iMac is performing within the design specs of 95w TDP, minimum 3.6 GHz (from our tests it’s sustaining at 3.8 GHz). As stated, turbo boost was always meant to be a bonus, not a guarantee otherwise Intel should change it where the i9 would have a minimum clock speed of 4.7 GHz instead. And require a higher TDP than 95 w. Since this isn’t the case, the iMac is performing within the specs of the processor that Intel set.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
2019 iMac - i5 vs i9? Cooling and Throttling?
- Thread starter Black Diesel
- Start date
- Sort by reaction score
But WHERE on the Intel spec sheet lists this? I don’t see their designs state a 4.7Ghz minimum clock speed or a TDP higher than 95w. Therefore, Apple’s iMac is performing within the design specs of 95w TDP, minimum 3.6 GHz (from our tests it’s sustaining at 3.8 GHz). As stated, turbo boost was always meant to be a bonus, not a guarantee otherwise Intel should change it where the i9 would have a minimum clock speed of 4.7 GHz instead. And require a higher TDP than 95 w. Since this isn’t the case, the iMac is performing within the specs of the processor that Intel set.
I'm glad quagmire has been posting more useful apples-to-apples testing in the iMac section. giggidy giggidy
Then Intel is to blame here. Force the i9 to require 200w+ TDP and set the minimum clock speed to 4.7Ghz if that is the “design” of this processor
???
Note: in the right environment
Turbo Boost is bonus as you say. Not the minimum base. I'm on your side here.
What would Steve do?
Run the chip, then probably crumble up the spec sheet, call it ********, and throw it in the trash.
In the past, turbo frequencies were obtainable within the TDP. The term TDP has gotten "loose," for lack of a better word. Ultimately, it's up to Apple to implement the chip accounting for actual test results they observe in real world testing, not a piece of paper they got from Intel.
What's not clear to me, is if their Cinebench runs are with the Noctua air cooler and 95W restricted. The cooler is not shown in this specific test. However, another testing spree released the same day shows the Noctua air cooler listed.
Searching for "95" there is some good discussion about power delivery, in where they claim "stock" settings were used. This is to assume unrestricted as would be expected of Gigabyte. This, of course accounts for the ~2000 Cinebench R15 run. If it's getting that at 95W, that's insanely great! Are we sure they sell their machine restricted? They shouldn't be showing benches of machines running unrestricted, then sell them restricted.
"Restricted" also doesn't need to imply 95watts. A particular vendor could easily find a sweet spot to balance thermals / acoustics / performance.
I’m not sure how this is a “proper comparison to the i9 iMac” (nor am I sure how you came up with US $3,200—I get US $3,785 for a Puget Spirit with 16GB RAM and 512GB PCI-E SSD, cheapest 8 GB video card, and wireless keyboard and mouse, and that’s with a 2560x1440 display and before adding in the US $333 cost of their three-year warranty, vs US $169 for AppleCare+), but everything else you wrote in this post strikes me as spot on.
(Mind you, I’d love a Mac tower with expansion slots and good cooling. Maybe we’ll get one when Apple finally ship their new Mac Pro.)
Last edited by a moderator:
3.8ghz IS turbo boost. I am pretty sure 3.8 > 3.6. How is this Apple’s fault if the Intel chip can not achieve 5 ghz at 95w TDP?
Most of us love how these iMac are coming up. The i9 seems to be a beast, not enough number on the i5, but seems good for a lighter home environment (I so much want the i9 for how good it is, but for what I do, not sure if it's worth stretching the extra £360 and maybe better paying the Vega for the occasional gaming I do)
Last edited by a moderator:
I don’t get it. Why don’t YOU just ask Puget? Go straight to the source.
You’ve got the most questions but not doing the most work to get the answers and expect others to spoon feed you.
[doublepost=1554422970][/doublepost]
Reviewers and actual owners are happy.
Tell us what you do and the regularly used apps so we can see which setup is best.
Last edited by a moderator:
To be honest is going to be mostly office work (handling big data excel sheets and such) and browsing, with the occasional video transcoding for ipad through handbracke (I don't even do that too often as vlc reproduces just fine), some coding in R/vba/python, running some scripts and such and some gaming.
I'm thinking that for me i5 +Vega = 512 SSD would be good as it's about my cash limit, but then keep on thinking if I would actually notice the i9, and if it would mean the iMac would last 2 more years (my current iMac is the 2011 27")
Correct. 3.8 is > than 3.6. Intel and Apple are both to blame, and it would make a very interesting court case with each side having valid arguments. TDP is the thermal design power. It's a number that is associated with any given processor, and rates the amount of cooling one should expect to dissipate under "normal" work loads, blah blah. Ultimately, each "95 Watt TDP" processor may consume more or less. 95 watts is not a hard limit, unless it is set by Apple, Dell, Asus, etc... Accordingly, the implementation of each will be different, as is highlighted with the 2019 iMac.
For the purposes of this thread and decision of purchasing one of these models, the i9 operates in the lower range of it's specified frequency potential, whereas the i5 operates within the top end, contrary to what Apple claims. Though, the i9 sports 16 threads, whereas the i5 has 6.
What this means for prospective buyers is up for them to decide. At first sight, the i9 was an obvious upgrade, especially as Apple was only charging $150 over the retail difference. That's very low for them. Though, it turns out the i9 iMac is heavily restricted and performs more like an i7 (8 thread) per AnandTech. If I had to pick between Vega and i9, I'd rather up to the Vega. But, this is just me...
Yes yes yes, even if only operating on the lower end, it is still getting massive better benchmark scores than the i5, so if someone needs the power, AND WE ALL KNOW WE ARE OVERPAYING, it's still a good option to buy, which is not throttling at all and working as intended by Apple (and even Intel). Let's just get back to constructive and useful arguments please, this is getting old.
One thing to consider with the Vega, is that the trend has been to offload more and more CPU load onto the GPU. So, the GPU also has the argument of adding usable lifespan.
BareFeats always has good benchmarks. I'd imagine they'll publish something soon with regards to the new iMac. May help shed some more light.
Hey, thanks for that one Bboble, it was honestly helpful. I will keep an eye on that bareFeats you mention to see if something comes up.
I checked delivery times if I order today and they are 11-15th April. I wont be back home till the 18th anyway so I Can wait a few days before placing down my order.
I would expect the BareFeats out any day now...they're usually fairly quick on new releases. Another thing to note, is that the CPU may still be socketed as was previous. (I cannot confirm this yet for 2019.) Though invasive, it would give you an upgrade path down the road.
Either way, you will be very pleased with the upgrade from 2011.
i9/580X and an eGPU down the road.
The i5 will not age well and stutter in games that pegs it. Most notable with 4 core / 4 thread CPU owners playing current games.
I had an RX580 eGPU for my 2011 MBP which put in extra life like when I put a SSD in.
Quite interesting.
The issue I find with the eGPU is that I don't really have any plans for another monitor (no room nor option to upgrade desk) and reviews say that eGPU don't work all that well with the internal monitor?
Do you think the 8 Core would really benefit for the type of work I do? or is it more due to the aging?
I don't see the conflict. On the page where Puget Systems listed the Cinebench R15 score of 2016 points on an i9-9900k, they did not state the *specific* test system or config or cooler or whether it's a commercial product. They only listed a motherboard. Nowhere did they state this was achieved on a specific Puget system with a certain configuration. At the bottom they list two system categories, one of which cannot even use the i9-9900k. For all we know it was a bench test to update customers on technology. Note the URL says "labs": https://www.pugetsystems.com/labs/articles/Cinema-4D-Intel-Core-i7-9700K-i9-9900K-Performance-1261/
Puget is a great company with a reputation for high quality and good support. I have a direct, personal email from them stating they confirmed with their production dept. they use the default 95W TDP on their Puget Spirit system. Most PC companies wouldn't even answer a question like that. If I was buying a Windows PC it would be from them.
But the main issue is neither Puget or Dell nor HP nor any other mainstream PC manufacturer apparently ships air-cooled business-class i9-9900k machines configured for unrestricted thermal mode. If they don't generally do this, that's no different than the i9 iMac.
This is how the situation currently appears. As more tests and comparisons are done on the i9 iMac we will gradually get a better picture.
Last edited:
eGPUs (at least from Nvidia) work well with the internal monitor in Boot Camp, but I’m told they lose some performance. (A high-end eGPU will still perform much better than the iMac’s Radeon Pro Vega 48, even when driving the internal monitor.)
In macOS, however, eGPU support for the internal monitor must be explicitly added by the application or game developer, otherwise an external monitor connected directly to the eGPU is your only option. Of all of Feral Interactive’s many macOS games, for example, only Rise of the Tomb Raider currently has eGPU support for the internal monitor.
(As far as I know, there’s no driver support at all for Nvidia GPUs in Mojave, not even from Nvidia, let alone Apple. In Boot Camp, AMD eGPUs can suffer from driver conflicts with internal AMD GPUs, but the unofficial and unsupported drivers from BootCampDrivers.com reportedly solve this—although they won’t work with the Vega 48 before June. But then Apple don’t officially support eGPUs in Boot Camp at all, so if you want to use one in Windows, you’re going to have to spend some time researching at eGPU.io and tinkering.)
If gaming is important to you and you don’t want to add an external monitor, then you should certainly get the internal Vega 48, even if it means getting the i5 instead of the i9.
Bear in mind also that while a Radeon VII in a Sonnet eGFX Breakaway Box 650, for example, will certainly blow a Radeon Pro Vega 48 out of the water, it will also set you back by US $1,100.
[doublepost=1554430995][/doublepost]Luke Miani today uploaded to YouTube a video review of the 2019 iMacs in which he covers many of the issues that we’ve been discussing in this thread, with detailed numbers from his testing.
The review has something for everyone here, enabling anyone who so wishes to cherry-pick and say, “See? I told you so!”
Last edited:
Haven't followed the eGPU scene in the last six months but back than it was ~25% avg performance loss to internal monitor.
I see the CPU more beneficial than a GPU for day to day use. I have to watch a video or edit to kick the GPU past 20% usage and game to keep it past 20%.
Whatever game you plan to play in the future will make use of the threads. These big games starting to utilize more cores & threads. CPU with no multi-thread will have higher variable FPS and more frame stutters as the future comes.
You can be stuck with the Vega 48 performance for the duration of the machine or invest in eGPU down the road. No right answer here.
You'll get a sizable savings going 580X to use towards eGPU than get one in the future when the need for more graphics arise. One thing is for certain, there will be better GPU's than now that will make it worth using even with a performance hit and take the heat away from the iMac internal. The latter was a benefit I didn't even think about until I notice it took more load to get the fans spinning higher in the MBP.
Check out my old Geekbench tests with and without eGPU:
6750M: https://browser.geekbench.com/v4/compute/2386312
RX580: https://browser.geekbench.com/v4/compute/2386309
Haha, that’s some upgrade!
By contrast, check out these new Geekbench tests with the 2019 iMac:
Not much difference! I’m guessing (hoping?) that’s simply because the brand-new Radeon VII driver in macOS 10.14.5 Beta (on which that system is running) has not yet undergone tuning.
If we look at all of the Macs running the Radeon VII in eGPUs and sort by score, we see that the Geekbench 4 Compute scores are higher in Windows than in macOS (which would seem to support my guess that there’s room for improvement in the current beta macOS drivers), but we also see that the Vega 48 in an iMac performs at a respectable 75% of the compute performance of the Radeon VII in an eGPU in Windows!
(Framerates in games, of course, are not limited only by raw compute power. I do not expect an iMac with a Vega 48 to give 75% of the gaming performance of a Radeon VII driving an external monitor.)
Their Spirit system—as well as the Design & Animation machine shown on the Cinebench page—is listed as having a "4.7–5.0 GHz Turbo," hence the confusion.
Furthermore, someone asked the specific question of power restriction, in where moderator Matt Bach from Puget stated: "We tend to test at stock settings since that is typically what we keep our workstations at." Another moderator, William George, later stated: "I haven't dug into the many settings available on the Gigabyte Z370 AORUS Gaming 5 that we have been using, but it must have power settings cranked up by default - since as Matt noted, we avoid changing settings from the defaults unless something is clearly wrong."
The original poster finished with this: "Just wanted to make sure that it was set to use the higher TDP it is made for, and not just the 95w.. but most importantly, that we know what it was tested woth so we can compare apples to apples
"Looks like Dell does a air-cooled 9900K machine.
Was digging around on Puget's site at all their benchmarks, articles, etc... (really great company from what I can discern) and stumbled upon their take on TDP:
My personal opinion - and I think it aligns pretty closely with our company policy, though I wouldn't want to speak for that without consulting other folks here - is that reaching the turbo speeds prescribed by the CPU manufacturer (either Intel or AMD) is not overclocking, so long as voltage is left at default settings.
In other words, I don't care about the TDP
We use CPU coolers which can handle heat far in excess of the TDP on the 9900K, and other models, even for extended periods of time. Because of that, what I want to see is processors reaching and maintaining the turbo clock speeds appropriate for however many cores are currently active. Thermal throttling as a last-ditch protection is fine and good (in case a fan fails, for example) but I don't want to see my processors throttling because of some artificial power draw limitation.
To me, actual overclocking is when you run part of a CPU (the core clock, the memory controller clock, etc) at a speed above what the manufacturer has rated it for. For example, using memory above 2666MHz on the 9900K - or setting it to run at 5.0GHz turbo across all cores, rather than when just 1-2 cores are under load. Overvolting is similar, but instead of changing clock speeds it involves increasing voltage to the CPU - and it is often required in order to enable overclocking to succeed. Both of those push a CPU beyond what the manufacturer has rated it for, though, whereas allowing higher wattage operation (at default clocks and voltages) isn't exceeding any performance specs, it is just allowing more heat to be generated. As long as that is handled responsibly - with a good CPU cooler and plenty of airflow through the system - then I think it is just fine and the "best practice" in my opinion.
I'm pretty ignorant. What does the compute score for a gpu represent in terms of video performance. I'm not a gamer. I do a lot of photo editing. Until recently, gpus didn't have a huge impact on most of that software. Most reviews said high end gpus provided little added benefit beyond a mid level card with 4gb. The trend seems to be toward AI. Adobe recently added an AI resolution enhancement tool to Lightroom and it seems to make much more use of the gpu. My guess is that gpu performance will become critical as some of the software evolves. I really had no thought of upgrading to a Vega, but after seeing some of the LR tests it's now kind of on my radar.