F5F Stay Refreshed Hardware Desktop The grumpy old guy is struggling to boost his 5800X processor.

The grumpy old guy is struggling to boost his 5800X processor.

The grumpy old guy is struggling to boost his 5800X processor.

Pages (2): 1 2 Next
C
ColinPlays19
Member
57
02-28-2020, 04:45 PM
#1
Hi there. Honestly, I’m not sure where to start from anymore. I’ve already tried pushing the clock speed higher, but the outcomes don’t make much sense. Used to feel confident with this kind of work, but I’ve owned a 4th gen i5 for about eight years now. I guess I fell behind Big Grin. My current configuration includes a Ryzen 5800X, an MSI X570 MAG Tomahawk, Wi-Fi Noctua NH-D14, Corsair Vengeance LPX 4000MHz, and a few other parts. Originally, my goal was to hit around 2000MHz base clock. Of course, I didn’t expect that much. If it doesn’t work, I’ll try lower frequencies like 1900 or 3800 MHz on the RAM and adjust the timings instead. While tweaking settings, I used OCCT for benchmarking, but I think it’s not giving me accurate data. Eventually, I went through a lot of adjustments, but they only made things worse. I noticed my CPU temperature stayed high at about 90°C, so I thought the cooling was the issue. At that time, I had some liquid metal inside, but I suspect the real problem was the case placement—sitting in a tight corner of my desk. In short, the best results came when the machine was moved out of that spot, using Curve Optimizer at -25, XMP profile, and a 1900 FCLK. That setup hit around 15592 multithread points in Cinebench. Initially, idle temps were in the low 30s, and under load they stayed below 80°C. I also experimented with CPU voltage offsets and found that lower CB scores consistently dropped. So I returned to my optimal setup (with saved profile), and now idle is between 40–50°C, while Cinebench peaks at about 85°C. It’s strange—every article I read about the perfect setup seems to ignore my specific situation. Even with 4000MHz RAM and a 1900 FCLK, it still outperforms 3800MHz and 1900 for certain tasks, despite having similar memory speeds. I’m not sure if there’s a one-size-fits-all answer, or if the right settings depend on subtle factors like voltage, voltage range, and even case design. I’m starting to feel frustrated because it seems like every adjustment I make only makes things worse. I’d really appreciate a step-by-step guide that explains exactly how each change affects the system. Also, I’m thinking about cutting a hole in the front of the case for a 140mm fan—those temperatures are really annoying.
C
ColinPlays19
02-28-2020, 04:45 PM #1

Hi there. Honestly, I’m not sure where to start from anymore. I’ve already tried pushing the clock speed higher, but the outcomes don’t make much sense. Used to feel confident with this kind of work, but I’ve owned a 4th gen i5 for about eight years now. I guess I fell behind Big Grin. My current configuration includes a Ryzen 5800X, an MSI X570 MAG Tomahawk, Wi-Fi Noctua NH-D14, Corsair Vengeance LPX 4000MHz, and a few other parts. Originally, my goal was to hit around 2000MHz base clock. Of course, I didn’t expect that much. If it doesn’t work, I’ll try lower frequencies like 1900 or 3800 MHz on the RAM and adjust the timings instead. While tweaking settings, I used OCCT for benchmarking, but I think it’s not giving me accurate data. Eventually, I went through a lot of adjustments, but they only made things worse. I noticed my CPU temperature stayed high at about 90°C, so I thought the cooling was the issue. At that time, I had some liquid metal inside, but I suspect the real problem was the case placement—sitting in a tight corner of my desk. In short, the best results came when the machine was moved out of that spot, using Curve Optimizer at -25, XMP profile, and a 1900 FCLK. That setup hit around 15592 multithread points in Cinebench. Initially, idle temps were in the low 30s, and under load they stayed below 80°C. I also experimented with CPU voltage offsets and found that lower CB scores consistently dropped. So I returned to my optimal setup (with saved profile), and now idle is between 40–50°C, while Cinebench peaks at about 85°C. It’s strange—every article I read about the perfect setup seems to ignore my specific situation. Even with 4000MHz RAM and a 1900 FCLK, it still outperforms 3800MHz and 1900 for certain tasks, despite having similar memory speeds. I’m not sure if there’s a one-size-fits-all answer, or if the right settings depend on subtle factors like voltage, voltage range, and even case design. I’m starting to feel frustrated because it seems like every adjustment I make only makes things worse. I’d really appreciate a step-by-step guide that explains exactly how each change affects the system. Also, I’m thinking about cutting a hole in the front of the case for a 140mm fan—those temperatures are really annoying.

C
Commando__
Senior Member
744
03-06-2020, 12:41 PM
#2
The temperatures are acceptable. 85°C during heavy use isn’t a big concern. I believe the issue here relates to different performance factors. Results depend on the context. Running async with increased memory speeds could improve scores in certain tests, but it might hurt other tasks. The aim of moderate overclocking is to boost performance for your specific tasks. When I tuned my 5900X, I adjusted settings that boosted multicore speed but slightly reduced single-core efficiency. For someone managing many projects, that trade-off makes sense. Multicore capability is crucial for my workflow. You should assess your workload types and prioritize optimization accordingly. Pursuing an arbitrary benchmark doesn’t matter much, especially without special cooling solutions.
C
Commando__
03-06-2020, 12:41 PM #2

The temperatures are acceptable. 85°C during heavy use isn’t a big concern. I believe the issue here relates to different performance factors. Results depend on the context. Running async with increased memory speeds could improve scores in certain tests, but it might hurt other tasks. The aim of moderate overclocking is to boost performance for your specific tasks. When I tuned my 5900X, I adjusted settings that boosted multicore speed but slightly reduced single-core efficiency. For someone managing many projects, that trade-off makes sense. Multicore capability is crucial for my workflow. You should assess your workload types and prioritize optimization accordingly. Pursuing an arbitrary benchmark doesn’t matter much, especially without special cooling solutions.

T
TheKroksBG
Member
209
03-06-2020, 04:55 PM
#3
I noticed that using an unlocked PBO on a single core paired with a small static boost up to 4600 MHz works well.
T
TheKroksBG
03-06-2020, 04:55 PM #3

I noticed that using an unlocked PBO on a single core paired with a small static boost up to 4600 MHz works well.

P
PRO__PVP
Member
64
03-06-2020, 06:09 PM
#4
P
PRO__PVP
03-06-2020, 06:09 PM #4

H
heyitsjackson
Member
170
03-06-2020, 10:59 PM
#5
Based on what I've seen, adjusting the CPU ratio might have locked it at maximum speed, which could increase power consumption.
H
heyitsjackson
03-06-2020, 10:59 PM #5

Based on what I've seen, adjusting the CPU ratio might have locked it at maximum speed, which could increase power consumption.

X
xEchoz
Member
208
03-13-2020, 03:53 AM
#6
It's not completely negative. You must identify the performance characteristics of the apps you use—whether they're mainly CPU-intensive or memory-intensive. If it's CPU-related, decide between single-core or multicore optimization. For memory issues, focus on bandwidth or latency. Make your choices carefully because they can restrict progress in other areas. For instance, if an app is memory-bound and very sensitive to delays, prioritize its timing parameters, even if it means reducing clock speed. After securing that, shift attention to the next most critical factor.
X
xEchoz
03-13-2020, 03:53 AM #6

It's not completely negative. You must identify the performance characteristics of the apps you use—whether they're mainly CPU-intensive or memory-intensive. If it's CPU-related, decide between single-core or multicore optimization. For memory issues, focus on bandwidth or latency. Make your choices carefully because they can restrict progress in other areas. For instance, if an app is memory-bound and very sensitive to delays, prioritize its timing parameters, even if it means reducing clock speed. After securing that, shift attention to the next most critical factor.

R
Redstoner137
Posting Freak
811
03-25-2020, 11:07 AM
#7
The motherboard features a capability to toggle between high-precision performance and a fixed overclock based on power consumption and temperature conditions.
R
Redstoner137
03-25-2020, 11:07 AM #7

The motherboard features a capability to toggle between high-precision performance and a fixed overclock based on power consumption and temperature conditions.

K
koen_maris
Junior Member
7
03-25-2020, 07:29 PM
#8
Honestly, I feel like I should stop working on this right now... It's getting really frustrating. This might not be worth the effort after all...
K
koen_maris
03-25-2020, 07:29 PM #8

Honestly, I feel like I should stop working on this right now... It's getting really frustrating. This might not be worth the effort after all...

F
FureaMC
Senior Member
564
03-28-2020, 06:19 AM
#9
Sure, just enable PBO, leave the limits off, and let the CPU operate naturally. Forcing a specific frequency will restrict its ability to adjust as needed. Concerning temperatures, the 5800X is the hottest among all Zen 3 CPUs. To boost performance, ensure your RAM is optimized with the best timings and MT/s settings. You might already have done that. Stay safe.
F
FureaMC
03-28-2020, 06:19 AM #9

Sure, just enable PBO, leave the limits off, and let the CPU operate naturally. Forcing a specific frequency will restrict its ability to adjust as needed. Concerning temperatures, the 5800X is the hottest among all Zen 3 CPUs. To boost performance, ensure your RAM is optimized with the best timings and MT/s settings. You might already have done that. Stay safe.

W
willdoespixel
Member
208
03-28-2020, 08:01 AM
#10
Stay focused on one task at a time. This chip is a single-core design, which causes it to overheat due to high transistor density. Unless you implement a full custom loop, performance will remain limited. Optimize timing by enabling PBO or increasing EDC and PPT limits—this should improve stability. Adjust memory voltage for tighter timing; higher voltages can help if your IMC is around 1.08V with XMP active. At 4000mt/s you’re already close to the limit, so stability may be the main concern. If needed, address it later. Generally, this rig runs very fast with just XMP enabled—expect minor fluctuations only.
W
willdoespixel
03-28-2020, 08:01 AM #10

Stay focused on one task at a time. This chip is a single-core design, which causes it to overheat due to high transistor density. Unless you implement a full custom loop, performance will remain limited. Optimize timing by enabling PBO or increasing EDC and PPT limits—this should improve stability. Adjust memory voltage for tighter timing; higher voltages can help if your IMC is around 1.08V with XMP active. At 4000mt/s you’re already close to the limit, so stability may be the main concern. If needed, address it later. Generally, this rig runs very fast with just XMP enabled—expect minor fluctuations only.

Pages (2): 1 2 Next