• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

lvl1 or lvl2 hardware prefetch in bios or both?

AMF

Joined
Apr 28, 2021
Messages
721 (0.48/day)
Location
USA
System Name MR T System price $2464
Processor Ryzen 9 3900x 325$ 4.725Ghz boost pbo +50+75 enhance .1875.-+offset auto voltage -.2500 offset soc
Motherboard ROG STRIX X570-E Gaming $ 125 open box
Cooling 280mm ARCTIC FREEZER II idle temps 25c cpu 28c cpu diode 20c vrm 25c motherboard b stock 60$
Memory 2x 16GB Dual rank G Skill Trident Z Neo F4-3600C16-16GTZNC @3800 1.405v 16-19-16-19-35-54-1T $96
Video Card(s) Asrock RX 6950XT Oc formula 705$ pcie4x16 2599 min 2699mhz 2799boost 1110v 2392mhz mem 19GB/s
Storage 500 GB WD SN850 PCI-E 4.0x4 85$ pcie 4.0x4
Display(s) Vizio M50QXM-K01 4k@120hz 2.5k 120hz 2k@240hz $448
Case Cougar Panzer Max G Full Tower Tempered Glass 150$
Audio Device(s) Vzio M51a 5.1 Soundbar Dolby Atmos goodwill 45$!!!!
Power Supply Seasonic SS-1250XM2 Bought in 2013 350$ 11 y/o hasnt lost any voltage or watts
Mouse Perixx eRGO wireless Trackball 25$ 2000dpi
Keyboard Corsair K57 50$
VR HMD Dont give a damn about vr
Software win 11 pro modded by me no ai no remote connections free hehe
Benchmark Scores 8188 multi 540 single cinebench r20
im tryingto figure what is the best setting. can anyone help?
 
Best left untouched.
 
Hardware prefetchers work well in workloads that traverse array and other regular data structures. The hardware prefetcher options are disabled by default and should be disabled when running applications that perform aggressive software prefetching or for workloads with limited cache.

This default setting is "Enabled". In some cases, setting this option to Disabled may improve performance. Users should only disable this option after performing application benchmarking to verify improved performance in their environment.

(source: google) :)
 
Last edited:
REASON? what does it actually do?

It has to do with the way the processor fetches data into it's cache, the reason it's best to leave it alone is because for one whatever effect it has it's likely imperceptible most of the time and secondly it's so application specific and nuanced that it's not even worth messing with.

I don't know why they even have this option available on consumer hardware, it's something only in a server/datacenter environment you'd even consider changing.
 
Disabled get's stability, Enabled get's higher performance per MHz.
You decide what you need :
1) Stability range is between 0 and "I don't know how many more MHz AFAIK no one checked", but you can be first ;)
2) Performance increase is also 0 to "I don't know how many more MHz AFAIK no one checked", but you can be first ;)
Nothing more is needed to know.
 
  • Like
Reactions: Jey
Disabled get's stability, Enabled get's higher performance per MHz.
You decide what you need :
1) Stability range is between 0 and "I don't know how many more MHz AFAIK no one checked", but you can be first ;)
2) Performance increase is also 0 to "I don't know how many more MHz AFAIK no one checked", but you can be first ;)
Nothing more is needed to know.
Is the stability software related ? (depending on what program / game your are using ?)
 
thanx guys yallwere helpful i thought as much about off being more stable
 
Got back with some weird analogy , its like counting on your instincts when doing some type of team work , if your team m8s are easy to predict (or you have synergy) you can do the job faster predicting where they will be at some point in time, you are going to be there before you see them going there and you can speed up the work done but if your team is harder to predict you will do the job worse because you will lose time going somewhere where there is no work to be done as your team m8 will be elsewhere (can be from 2x to whatever slower depending on work) , its pure luck in this situation to increase the work done , statistically will be worse, i think is the same situation with prefetch.
Edit: to get back to my own question , my guess is that is depending on the software used. (dont take it as valid , its just my logic).

Edit2: i think my analogy was bad in some way as you have to predict the work not the workers , but the idea is the same , you send people to do the work to a location where you think the work has to be done and if there is no work there you lose time (performance) ... etc.. i could had been more accurate (even this is not that accurate) but i was only trying to explain the logic as i see it on fast forward.
 
Last edited:
  • Like
Reactions: AMF
Back
Top