Overclock.net › Forums › AMD › AMD - General › Where are MOFSETS, and which is better to OC, FSB or Multiplier??
New Posts  All Forums:Forum Nav:

Where are MOFSETS, and which is better to OC, FSB or Multiplier?? - Page 4

post #31 of 48
Quote:
Originally Posted by gcampton View Post
14 x 300 = 4200
15 x 280 = 4200
16 x 262 = 4192
17 x 247 = 4199
18 x 233 = 4194
19 x 221 = 4199
20 x 210 = 4200

What should I pick? I can guarantee they will all be stable.
(CPU Multi) * (FSB) = (CPU Freq)
(CPU Multi) / (Memory Divider) = (Divisor Ratio)
(CPU Freq) / (Divisor Ratio) = (RAM MHz) (* 2 = DDR MHz)
(NB Multi) * (FSB) = (NB Freq)

So back to your ram speed question...

Looks like your working with strictly the CPU multiplier and FSB to achieve the CPU Freq... were I am coming from is that in (my case as an example)

16.5 Multiplier (250 fsb) = 4125 Freq
16.5 / (1:1/3) = 4.95 Div Ratio
4125 / 4.95 = 833.33~ Ram MHz (x2 its ddr) = 1666.7~ DDR Mhz

Where that ram speed comes into play is the 1666.7~ DDR Mhz speed, or rather, a post at that ram speed without issues on 1600mhz ram. So meh ram is slightly OC'd. Having 2k rated ram opens some possibilities as to what you may be able to get away with on the fsb with your available memory dividers as opposed to relying on the multiplier to get you to your cpu freq.

When I do an OC' I normally make a little chart and than map out what the mobo/memory/chip will tolerate (a post without a cmos reset) and then come up with some combinations that give me the biggest boost across the board, while staying stable. That normally means a slew of benchmarking, but as I was saying, LinX (intel burn test) will poo its pants quick if your not stable... in that case, you may map out some voltages that allow you to get past that instability, and watching temps... core temps are fine for cpu, something like Speedfan will read the thermistor (if installed) on the mobo under the chip, and give you your nb temps as well... although many of the thermistor numbers are erroneous, they tend to be true to themselves, which means seeing what your results are and plotting the Delta T's yourself.

As far as 'tightening up' these settings, finding that 'odd combination' that just works, or lends itself to a higher G.flop rating through multiple test without losses. 5 test within a couple hundreths of a G.flop is better than 1~2 G.flop variance on the same five test.
 
Yukikaze
(18 items)
 
 
CPUMotherboardGraphicsRAM
AMD Ryzen 1800X Asus C6 Hero AMD / ATI Reference AC VEGA 64cu - EK-FC-WB G.Skill DDR4 3200 2x8 Single Samsung B.Dies 
Hard DriveCoolingCoolingOS
Samsung EVO 960 NVME 250G Water Air Winblows 10c 
MonitorKeyboardPowerCase
Samsung CF791 34" 3440x1440 x100hz 48-100hz Fre... Logitech G15 TPX-775M NZXT Phantom "Original" 
MouseMouse PadAudioOther
G700 Generic Black Neoprene Supreme FX S1220 Onboard Aerocool Touch 2100 
OtherOther
Revoltec CCFL 30cm Twin UV Rev 2 Revoltec CCFL 30cm Twin UV Rev 1 
CPUMotherboardGraphicsRAM
Phenom II X4 965 be MSI NF980-G65 MSI R9 280X (2x) G.Skill F3-12800CL9D-4GBRL 
Hard DriveOptical DriveCoolingOS
SSD DVD-RW 24x Custom water Windows 7 (build 7600) 64bit 
MonitorKeyboardPowerCase
B2230 Logitech G15 w/ LCD TPX-775M NZXT Phantom White 
MouseMouse PadAudioAudio
Logitech MX Revolution a black square SB X-Fi Titanium Sharkoon XTatic 
AudioOther
Logitech THX 5.1 Original G15 / Logitech G700 
  hide details  
Reply
 
Yukikaze
(18 items)
 
 
CPUMotherboardGraphicsRAM
AMD Ryzen 1800X Asus C6 Hero AMD / ATI Reference AC VEGA 64cu - EK-FC-WB G.Skill DDR4 3200 2x8 Single Samsung B.Dies 
Hard DriveCoolingCoolingOS
Samsung EVO 960 NVME 250G Water Air Winblows 10c 
MonitorKeyboardPowerCase
Samsung CF791 34" 3440x1440 x100hz 48-100hz Fre... Logitech G15 TPX-775M NZXT Phantom "Original" 
MouseMouse PadAudioOther
G700 Generic Black Neoprene Supreme FX S1220 Onboard Aerocool Touch 2100 
OtherOther
Revoltec CCFL 30cm Twin UV Rev 2 Revoltec CCFL 30cm Twin UV Rev 1 
CPUMotherboardGraphicsRAM
Phenom II X4 965 be MSI NF980-G65 MSI R9 280X (2x) G.Skill F3-12800CL9D-4GBRL 
Hard DriveOptical DriveCoolingOS
SSD DVD-RW 24x Custom water Windows 7 (build 7600) 64bit 
MonitorKeyboardPowerCase
B2230 Logitech G15 w/ LCD TPX-775M NZXT Phantom White 
MouseMouse PadAudioAudio
Logitech MX Revolution a black square SB X-Fi Titanium Sharkoon XTatic 
AudioOther
Logitech THX 5.1 Original G15 / Logitech G700 
  hide details  
Reply
post #32 of 48
Quote:
Originally Posted by gcampton View Post
Also what I said earlier about ram, it AMD really capped to 1333Hmz ram?

Because I purposefully bought 2000Mhz ram thinking it would be better. Cost me around $150 for 4gb(CL9). when I could have got 8gb 1333mhz ram @ CL8, and about half the price of the other...
Pfffftt
You can get 1600Mhz RAM for even cheaper than all that.... yes, in Australia.
AMD has no problem with 1600Mhz RAM, in fact there's even an x8 multiplier that allows you to run 1600Mhz RAM. Ignore anything motherboards say about RAM speed support.

mosfets are small flat black squares, behind the big black squares called the chokes.
post #33 of 48
Quote:
Originally Posted by Maximus Desces Meridius View Post
Coretemp is my favorite.

Why does it suck?
Well, in my case, it doesn't seem to play well with AMD CPUs. I cannot seem to get any useful temp information from it or RealTemp. Just a lot of zeros. or odd negative numbers. Seems to be an issue with most temp monitoring software and Phenom II CPUs; from my research. I haven't tried HWMonitor yet, so I don't know how well it works.
Gormenghast
(16 items)
 
  
CPUMotherboardGraphicsRAM
AMD Phenom II X4 970 Black Edition Deneb ASUS M4N98TD EVO Asus ENGTX460TOP768 dual SLI Crucial Ballistix DDR3 PC310600 
Hard DriveOptical DriveCoolingOS
HGST HMS5C4040ALE640 Plextor PX-810SA Corsair Hydro H70 Microsoft Windows 7 Ultimate 
MonitorKeyboardPowerCase
ViewSonic VX2240w - dual Microsoft Natural Ergo 4000 Thermaltake Toughpower 700W and Thermaltake Pow... Thermaltake Armour w/250mm side fan 
MouseMouse PadAudioOther
Logitech Trackman Marble FX Cyborg V.5 Onboard Cyborg R.A.T. 7 gaming mouse 
  hide details  
Reply
Gormenghast
(16 items)
 
  
CPUMotherboardGraphicsRAM
AMD Phenom II X4 970 Black Edition Deneb ASUS M4N98TD EVO Asus ENGTX460TOP768 dual SLI Crucial Ballistix DDR3 PC310600 
Hard DriveOptical DriveCoolingOS
HGST HMS5C4040ALE640 Plextor PX-810SA Corsair Hydro H70 Microsoft Windows 7 Ultimate 
MonitorKeyboardPowerCase
ViewSonic VX2240w - dual Microsoft Natural Ergo 4000 Thermaltake Toughpower 700W and Thermaltake Pow... Thermaltake Armour w/250mm side fan 
MouseMouse PadAudioOther
Logitech Trackman Marble FX Cyborg V.5 Onboard Cyborg R.A.T. 7 gaming mouse 
  hide details  
Reply
post #34 of 48
Noob question here... I know people have their own methods, but so long as the CPU freq, NB freq, Mem freq and timings are the same then you will get the same performance no matter if you get there by FSB or by each multi separately, RIGHT?
Or does one way or the other give you some performance boost?
post #35 of 48
Thread Starter 
Quote:
Originally Posted by Maximus Desces Meridius View Post
Coretemp is my favorite.


Why does it suck?
coretemp didn't give me CPU temps (NB, mobo), only core temps. Which are pretty darn useless considering I have an unlocked core. 0c | 0c | 0c

HWmonitor did, but it was called something funny. Like Td02 or something.
Really thinking about purchasing AIDA64.
I dunno, bit of a rip-off considering you should be able to get all its functionality for free.
post #36 of 48
Thread Starter 
Quote:
Originally Posted by Soggysilicon View Post
(CPU Multi) * (FSB) = (CPU Freq)
(CPU Multi) / (Memory Divider) = (Divisor Ratio)
(CPU Freq) / (Divisor Ratio) = (RAM MHz) (* 2 = DDR MHz)
(NB Multi) * (FSB) = (NB Freq)

So back to your ram speed question...

Looks like your working with strictly the CPU multiplier and FSB to achieve the CPU Freq... were I am coming from is that in (my case as an example)

16.5 Multiplier (250 fsb) = 4125 Freq
16.5 / (1:1/3) = 4.95 Div Ratio
4125 / 4.95 = 833.33~ Ram MHz (x2 its ddr) = 1666.7~ DDR Mhz
That's brilliant thanks , now I just have to figure out what it all means
post #37 of 48
Thread Starter 
1 more question:

HT Link Width: Options -> Auto -> 8bit -> 16bit
HT Link Frequency: Options -> Auto -> x1~x10

What is this in relation to? What is this exactly? because the manual doesn't describe what it does, and wikipedia says it's a replacement for FSB, PCIe, and many other things, cause it's awesome and can support DDR.... WOW!


edit:
OK I lied, another question. I can change memory clock to:
x4.00
x5.33
x6.66
x8.00
Pretty much same as CPU multiplier. should I try to max it around 2ghz or just keep it around 1666Mhz?

I just looked in bios to change ram latency, and OMG! so many options. literally I have about 20-25 different ram over clocking options. None of which make any sense, but I managed to get 8-8-8-22 working fine, burn tested, I perhaps might reduce it further but first will see how it posts with 17x240.
Edited by gcampton - 2/21/11 at 7:10am
post #38 of 48
Thread Starter 
Quote:
Originally Posted by Soggysilicon View Post
As far as 'tightening up' these settings, finding that 'odd combination' that just works, or lends itself to a higher G.flop rating through multiple test without losses. 5 test within a couple hundreths of a G.flop is better than 1~2 G.flop variance on the same five test.
Oh I got the impression from reading the "readme" that it's the result that must remain exactly the same for all tests. Why is GFlops important?

Not that I have had any seconds on gflops. mostly all just tenths of second.
Time (s) --- Speed (GFlops) --- Result
45.828 ----- 19.5052 ----- 3.456245e-002
45.288 ----- 19.7380 ----- 3.456245e-002
45.356 ----- 19.7082 ----- 3.456245e-002
45.154 ----- 19.7962 ----- 3.456245e-002
45.157 ----- 19.7948 ----- 3.456245e-002

that test was with ram latency @ 8CL was about 1580hmz.
Multi @ 17x (norm 16) (approx 3740 Mhz)
NB @ 12x (norm 10 @ 2Ghz) (approx 2640 Mhz)
HT @ 12x (norm 10 @ 2Ghz) (approx 2640 Mhz)
and FSB @ 220
thought I'd test 3.7 see how it went, now to up to 3.9.
Edited by gcampton - 2/21/11 at 7:47am
post #39 of 48
Thread Starter 
Seems I have been abandoned so I'll continue chatting to myself at least.

I reduced NB and HT link back down to x11, because was getting some problems with both being up around 2600, prefers to sit on 2400 on lower. So revision:

CPU Multi x18 @ 4050
FSB x 225
NB x11 @ 2475
HT x11 @ 2475
RAM Speed approx 1520
8 - 8 - 8 - 22
third core unlocked.
VCORE Volt @ 1.07, 1.42 load
DDR @ 1.5
3.3V @ 3.31
5V @ 5.05
12V @ 12.16
VBAT (1.42 min) 3.28

temps:
mobo 36c
CPU 33c, 49c load.
tmpin2 (dunno what it is) 40c, 52c load.
core's
post #40 of 48
adjust your ht link as close as possible to 2000. do you have cool&quiet turned on??
Red Dragon
(16 items)
 
  
CPUMotherboardGraphicsRAM
Amd Phenom II X4 960T @ 4.2Ghz @ 1.475v Asus Sabertooth 990FX @ NB 3.2Ghz @ 1.35v Sapphire ATI 5870 Vapor-X Corsair Dom 4GB DDR3 1600MHz 7-7-7-21-1t @ 1.65v 
Hard DriveCoolingOSMonitor
Samsung F3 (1+1)Tb Venomous-X + Scythe GT Window 7 Ultimate 32bit Samsung P2770H 27' 
KeyboardPowerCaseMouse
Razer Tarantula Corsair TX650W HAF 922 Logitech G9X  
Mouse PadAudioAudio
Razer Goliathus Altec Lansing VS4621 Roccat Kave 
  hide details  
Reply
Red Dragon
(16 items)
 
  
CPUMotherboardGraphicsRAM
Amd Phenom II X4 960T @ 4.2Ghz @ 1.475v Asus Sabertooth 990FX @ NB 3.2Ghz @ 1.35v Sapphire ATI 5870 Vapor-X Corsair Dom 4GB DDR3 1600MHz 7-7-7-21-1t @ 1.65v 
Hard DriveCoolingOSMonitor
Samsung F3 (1+1)Tb Venomous-X + Scythe GT Window 7 Ultimate 32bit Samsung P2770H 27' 
KeyboardPowerCaseMouse
Razer Tarantula Corsair TX650W HAF 922 Logitech G9X  
Mouse PadAudioAudio
Razer Goliathus Altec Lansing VS4621 Roccat Kave 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: AMD - General
Overclock.net › Forums › AMD › AMD - General › Where are MOFSETS, and which is better to OC, FSB or Multiplier??