Friday, July 17th 2020

AMD Ryzen 7 PRO 4750G Geekbenched, Gets Close to 3700X-level Performance

by
btarunr
 Discuss (12 Comments)
AMD's top upcoming Socket AM4 desktop APU, the Ryzen 7 PRO 4750G, was put through Geekbench 5, as discovered by TUM_APISAK. The processor produced performance figures in the league of the popular Ryzen 7 3700X desktop processor. Both are 8-core/16-thread processors based on the "Zen 2" microarchitecture, but while the 3700X has additional L3 cache and added power budget for the CPU cores (as the processor completely lacks an iGPU); the PRO 4750G offers a Radeon Vega 8 iGPU with its engine clock above 2.00 GHz. Both chips were compared on Geekbench 5.2.2.

The single-core performance of both the PRO 4750G and 3700X are similar, with the PRO 4750G scoring 1239 points, and the 3700X scoring 1266 points. The 3700X has a slight upper hand with multi-core performance, with 9151 points compared to 8228 points of the PRO 4750G. This is attributable to the 3700X enjoying four times the L3 cache size. The Ryzen 7 PRO 4750G is expected to be the top desktop SKU based on the 7 nm "Renoir" silicon that features eight "Zen 2" CPU cores, and an iGPU based on the "Vega" graphics architecture, featuring 8 NGCUs amounting to 512 stream processors. The processor features AMD PRO feature-set that make it fit for use in commercial desktops in large business environments.
Source: TUM_APISAK (Twitter)

Related News

Add your own comment

12 Comments on AMD Ryzen 7 PRO 4750G Geekbenched, Gets Close to 3700X-level Performance

#1
ratirt
that is great for a 40 watt processor in a desktop with an iGPU almost matching the 3700X desktop processor.
Posted on Reply
#2
Chrispy_
ratirtthat is great for a 40 watt processor in a desktop with an iGPU almost matching the 3700X desktop processor.
Would it surprise you to hear that the 3700X running at stock in a board with AMD-recommended PPT, TDC, EDC limits will only pull around 40-50W in most loads? The 65W TDP is generally a result of heavy voltage boost for higher single-core clocks, and at 'stock' PBO settings and power limits, Zen2 actually runs at lower power on more modest multi-core voltages and clocks typical of most real-world use cases.

Peak boost speeds are nice to have with PBO+ on an overclocking-focused motherboard, but Zen2 will get 95% of the way there at very tame voltages with bone-stock settings. Yes, a 3950X will POST and boot on a cheapo, garbage-tier $60 B450 board with no VRM heatsinks. You're not getting the performance potential you paid for, but it won't melt anything nor will it invalidate any warranties - the board is designed (and warranty must be honoured) to manage 105W CPUs and the CPU is designed to run in a board that only delivers 105W. Don't forget, a modest old Ryzen5 1600X is a 95W chip.
Posted on Reply
#3
Imsochobo
Chrispy_Would it surprise you to hear that the 3700X running at stock in a board with AMD-recommended PPT, TDC, EDC limits will only pull around 40-50W in most loads? The 65W TDP is generally a result of heavy voltage boost for higher single-core clocks, and at 'stock' PBO settings and power limits, Zen2 actually runs at lower power on more modest multi-core voltages and clocks typical of most real-world use cases.

Peak boost speeds are nice to have with PBO+ on an overclocking-focused motherboard, but Zen2 will get 95% of the way there at very tame voltages with bone-stock settings. Yes, a 3950X will POST and boot on a cheapo, garbage-tier $60 B450 board with no VRM heatsinks. You're not getting the performance potential you paid for, but it won't melt anything nor will it invalidate any warranties - the board is designed (and warranty must be honoured) to manage 105W CPUs and the CPU is designed to run in a board that only delivers 105W. Don't forget, a modest old Ryzen5 1600X is a 95W chip.
They run 77W in amd recommended no PBO.
in OEM machines I see more 75W.
Posted on Reply
#4
Chrispy_
ImsochoboThey run 77W in amd recommended no PBO.
in OEM machines I see more 75W.
That's full system draw from the wall socket, right? Mine's drawing 103W from the wall right now and Ryzen Master is claiming 61W of that is the CPU. I have four drives, 8 fans and an idle GPU running off a Gold PSU with ~10% efficiency losses, so those numbers check out to me.

Edit - using Furmark's CPU burner it's very quick to get more numbers, not 100% sure how good CPU burner is as a real-world example but it does provide a 100% load to a specified number of threads nice and easily:
  • 61W - 16 threads, 4.09GHz
  • 55W - 12 threads, 4.12GHz
  • 47W - 8 threads, 4.17GHz
  • 36W - 4 threads, 4.21GHz
  • 28W - 1 thread, 4.36GHz
These are numbers coming from Ryzen Master, FWIW and that's why my original comment was "40-50W" since I would imagine that typical workloads of chrome and a few applications use between 5-10 threads. This thing is almost never at 0 or 100% CPU - task manager seems to put it in the 3-70% range most of the time.
Posted on Reply
#5
ratirt
Chrispy_Would it surprise you to hear that the 3700X running at stock in a board with AMD-recommended PPT, TDC, EDC limits will only pull around 40-50W in most loads? The 65W TDP is generally a result of heavy voltage boost for higher single-core clocks, and at 'stock' PBO settings and power limits, Zen2 actually runs at lower power on more modest multi-core voltages and clocks typical of most real-world use cases.

Peak boost speeds are nice to have with PBO+ on an overclocking-focused motherboard, but Zen2 will get 95% of the way there at very tame voltages with bone-stock settings. Yes, a 3950X will POST and boot on a cheapo, garbage-tier $60 B450 board with no VRM heatsinks. You're not getting the performance potential you paid for, but it won't melt anything nor will it invalidate any warranties - the board is designed (and warranty must be honoured) to manage 105W CPUs and the CPU is designed to run in a board that only delivers 105W. Don't forget, a modest old Ryzen5 1600X is a 95W chip.
It doesn't surprise me but we are comparing laptop CPU to a desktop CPU and it's still impressive that these two have almost identical performance but are in different segments. You have to admit, AMD Ryzens are really good chips for what they can do and the power they consume.
Posted on Reply
#6
Chrispy_
ratirtIt doesn't surprise me but we are comparing laptop CPU to a desktop CPU and it's still impressive that these two have almost identical performance but are in different segments. You have to admit, AMD Ryzens are really good chips for what they can do and the power they consume.
Yep. Like all AMD things though, the desktop variants are pushed right to the silicon and power limits. If you dial them back a bit you get huge efficiency gains.

The other thing to remember about these is that the whole die is now the ultra-efficient 7nm process. Desktop 3000-series still have to feed a less-efficient 12/14nm IO die too.
Posted on Reply
#7
LabRat 891
Chrispy_Yep. Like all AMD things though, the desktop variants are pushed right to the silicon and power limits. If you dial them back a bit you get huge efficiency gains.

The other thing to remember about these is that the whole die is now the ultra-efficient 7nm process. Desktop 3000-series still have to feed a less-efficient 12/14nm IO die too.
We're also going to be seeing considerable latency improvement. I wouldn't at all be surprised to see Renoir Desktop APUs beating their Zen2 R3000 counterpart when both are pushed to 1800IF/3600DDR4 with pretty much no other tweaks.
I'd love a 6c/12t or 8c/16t Renoir. Would put a massive heatpipe cooler on it, set 1800IF and up the TDP limits.
Don't be surprised to see people making tools to disable the iGPU and allot max TDP all to the CPU. The newer silicon+no seperate I/O die should make for some good performing (and clocking) silicon if kept cool enough. Even no tweaking, the cooler you keep Ryzen the more free perf you get. Same with ARM chips in mobile devices, now that I think of it.
Posted on Reply
#8
tabascosauz
Chrispy_Would it surprise you to hear that the 3700X running at stock in a board with AMD-recommended PPT, TDC, EDC limits will only pull around 40-50W in most loads? The 65W TDP is generally a result of heavy voltage boost for higher single-core clocks, and at 'stock' PBO settings and power limits, Zen2 actually runs at lower power on more modest multi-core voltages and clocks typical of most real-world use cases.

Peak boost speeds are nice to have with PBO+ on an overclocking-focused motherboard, but Zen2 will get 95% of the way there at very tame voltages with bone-stock settings. Yes, a 3950X will POST and boot on a cheapo, garbage-tier $60 B450 board with no VRM heatsinks. You're not getting the performance potential you paid for, but it won't melt anything nor will it invalidate any warranties - the board is designed (and warranty must be honoured) to manage 105W CPUs and the CPU is designed to run in a board that only delivers 105W. Don't forget, a modest old Ryzen5 1600X is a 95W chip.
That's demonstrably false. Stock PPT is 88W, and in common stress tests that aren't P95 (which still obeys PPT at stock) will pull up to that without PBO. That's CPU power draw over SMU, not total system power. Hello? Your tweaked 3700X =! everyone else's stock 3700X.

Even after only playing a couple of recent games, mine is up to 65-75W package power draw on two different boards, one with underreported power draw and one that's roughly accurate. Both when stock, no PBO, on 1.0.0.4B and v2 1.0.0.2.

2 cores utilizing 1.5V to get to 4.4GHz momentarily isn't pulling 60W :wtf: also, a 105W 3950X allows up to 142W PPT and 140A EDC at stock. I know a ton of barrel scraping AM4 boards that either are realistically rated for far less than 140A output for any period of time, and/or will by design massively throttle their power delivery to the tune of 40%+.

If Renoir's PPT is set correspondingly 20-25W lower as its TDP is, it reflects the fact that its I/O die doesn't constantly draw 13W of power like Matisse does because Renoir can dynamically scale IF and integrates its I/O, as well as the natural process improvements of N7/N7P since July 2019. This process at launch was NOT ready for the clocks that AMD asked for.
Posted on Reply
#10
Chrispy_
tabascosauzHello? Your tweaked 3700X =! everyone else's stock 3700X.
Ummm...
At what point did you read that I'd 'tweaked' my 3700X?
This entire discussion - at least my part in it - has explicitly been talking about bone-stock settings.
I guess my BIOS still has the RAM timings from 1usmus instead of XMP timings, but it's still the same Patriot Viper DDR4-3600 running at 3600MHz
Posted on Reply
#11
Imsochobo
Chrispy_That's full system draw from the wall socket, right? Mine's drawing 103W from the wall right now and Ryzen Master is claiming 61W of that is the CPU. I have four drives, 8 fans and an idle GPU running off a Gold PSU with ~10% efficiency losses, so those numbers check out to me.

Edit - using Furmark's CPU burner it's very quick to get more numbers, not 100% sure how good CPU burner is as a real-world example but it does provide a 100% load to a specified number of threads nice and easily:
  • 61W - 16 threads, 4.09GHz
  • 55W - 12 threads, 4.12GHz
  • 47W - 8 threads, 4.17GHz
  • 36W - 4 threads, 4.21GHz
  • 28W - 1 thread, 4.36GHz
These are numbers coming from Ryzen Master, FWIW and that's why my original comment was "40-50W" since I would imagine that typical workloads of chrome and a few applications use between 5-10 threads. This thing is almost never at 0 or 100% CPU - task manager seems to put it in the 3-70% range most of the time.
SOC power draw is what I specified.
What the chip is using in total with memory, northbridge,uncore whatever you name it and is measured with an hp prebuilt.

worth noting some motherboards enable eco mode by default for some reason which locks it to 65W and there isn't much loss in performance, 100mhz or so..
Posted on Reply
Add your own comment
Jun 10th, 2024 16:03 EDT change timezone

Latest GPU Drivers

New Forum Posts

Popular Reviews

Controversial News Posts