Overclocking Sandbox: Tbred B DLT3C 1700+ and Beyond

What about the new SLK 800 U , which is a bolt on (screws on) would that have better contact then the clip on?

Between the SLK-800A and SLK-800U, I would go with the SLK-800U for few $ more because

1. SLK-800 has a 550 g mass, it is 65 g more than the SK7, the bolt will hold it onto the mb and contact the CPU more securely.

2. The SLK-800U can be used for P4 also.

lol p4 i hate them so.... lol

but yea thanx for the info man

That's just nuts. 2.5GHz on air has been done before, but your temps are awesome, especially considering the FSB and voltage you're running at.

Congratz

lol true.

^bump^ for the nice OC

god bless ny temps, i cant say the same here in mexico

I'm amb26c/sys42c/cpu52c right now under full FAH3 load @1.65v/200/2200.

It gets over 30c in my living room durring the summer, so I'll not be able to hold my O/C for much longer.

I'll have to drop my voltage to reduce the heating soon

30C is fine, it work roast out

We should arrange a lan oc party in the south pole, or move to the south pole for oc during the summer time in the northern hemisphere.


This is summary of what I learned and used in overclocking a Tbred B 1700+ DLT3C click link. Probably most of you already know about these techniques. If you think there are something missing or incomplete or incorrect, please post. But I think this may help some new comers to get start, ...

Summary for overclocking CPU and FSB

When overclocking a system, there are two major components.

1. CPU clock frequency
2. FSB clock frequency which is related to memory bus clock freq and ras/cas timing

The two can be optimized separately and resolved by the CPU multiplier (to the first degree).

First recent motherboard such as nforce2 can lock PCI bus and AGP bus speed, so assume PCI is locked at 33 MHz, so don't have to worry about HD and video card AGP bus running out of spec. For older mb whose FSB is running at a fsbCI ratio of 4:1 or 5:1, the max FSB is limited due to the PCI frequency should not exceed its spec too much. Usually 38 MHz is considered safe. E.g. if your mb only have a max 5:1 fsbCI multiplier, if max PCI is 38 MHz, you FSB is limited to 38x5 = 190 MHz.

1. CPU clock frequency: For a given CPU, it can be overclocked to its max frequency regardless of FSB speed, whether it is only running at 133 MHz or 166 MHz or overclock to 220-250 MHz (for the newer mb). So even you have to keep an older mb (that can run 133 or 166 FSB) for a while, don't get discouraged and can still overclock a powerful CPU such as the Tbred B 1700+ DLT3C or 2100+. The multiplier in the bios for AMD CPU will resolve that by

FSB x multiplier = CPU frequency

Assuming you have an unlocked CPU such as Tbred, e.g. if FSB = 200 MHz, you can set multiplier to 11.5 and so CPU, such as Tbred B 1700+ DLT3C or Barton 2500, can run at 200 x 11.5 = 2.3 GHz. Or if multiplier at 12, CPU would run at 2.4 GHz, or at 12.5, CPU would run at 2.5 GHz, etc, etc. Side track: some mb may cap the multiplier at 12.5 and you may need a mod to get to higher multiplier than 12.5. If needed you can find such info easily by a search.

CPU max overclocking frequency on air is mainly determined by
- the heat sink and fan (on CPU temp)
- Vcore and its stability (PSU rail and Vcore regulator cooling)
to the first degree, case and system temp secondary.

So for any system, assuming on air, if you have good heat sink such as SK7 (economical) or SLK-800U/900U and variable speed fan with speed 2500-5000 rpm that can push air up to 75 CFM (e.g. Thermaltake Smart Fan II) through the heat sink. Vantec Tornado fan has 84 CFM fan can cool even better, but a resistor mod in series w/ the fan is recommended so the fan speed/noise can be controlled. (Note that SLK-900U does not fit well physically on an A7N8X). The die temperature, such as Tbred B, should then be able to be kept under 50C at high level of oc. 40 C is even a more ideal temperature for oc frequency increase per Vcore increase (I estimated 100-140 MHz/100 mV for Tbred B 1700+ DLT3C). Always the lower the temperature the better, that is why water, thermoelectric, phase change, ... cooling come in. CPU fan speed can be lowered to between 3000-3500 rpm and the noise level is generally considered acceptable for 24/7 usage. AS3 or ceramique should help also.

Temperature is important not just because high temperature could damage the CPU (max temp from spec is actually much higher, 85-90 C for Tbred/Barton), more importantly lower temperature results in lower CPU leakage current which heats up the CPU drastically above certain temperature threshold (around 50 C), above which heat and current will positively feeding each other and the CPU will lose stability in oc'ing. Lower temperature means the chip can run faster, estimated (rule of thumb) around 0.4%/C for Tbred B/Barton. I did some calculation on various HSF, the difference between a HSF with 0.3 C/W thermal resistance (such as Vantec aeroflow, Volcano 7+) can be as much as 200-300 MHz lower compared to a SLK-800(U) (0.23 C/W). And with a HSF with 0.36 C/W (such as Volcano 11), the difference can be as much as 300-400 MHz+ lower, compared to SLK-800(U). Also I estimated that an increase in ambient temperature by 5 C (e.g. summer), there would be a 5-7% drop in max CPU oc frequency. Details are in another thread.

Good PSU delivers stable CPU Vcore (within +- 50 mV (AMD DC tolerance spec) at full Prime95 load), given the high active current surge at max oc within a very short period of time (possibly few CPU cycles). For high oc to 2.5+ GHz at 1.9V Vcore may require good PSU to source enough active Icore current so that the Vcore line, and the PSU 5V or 12V lines (depends on motherboard) that generates Vcore are stable (i.e. small line fluctuation). A7N8X, 8RDA+ uses 5V, NF7-S, 8RDA3+ uses 12V. As an example, a CPU running at 2.5 GHz 1.9V consumes 2.7 times the active power, and draws 2.1 times the active current of that running at 1.5 GHz at 1.5V.

I found that when choosing PSU, a major consideration should be shifted to current rating (on the PSU line that generates Vcore) for high level of overclocing, and not on the total power (watt) rating that most people seem to emphasize on and pay attention to when looking at PSU.

E.g. I did an estimate for a system with a Tbred B 1700+ DLT3C oc to 2.5 GHz 1.9 V, w/ video card such as 9700 pro, 2 HD's, DVD, CDRW, 1 CPU fan, 4 case fans, 2 memory sticks, ... for a nforce2 mb (such as NF7-S, 8RDA3+) which use 12 V to generate Vcore, the total 12V current requirement = 23.4 A +-. Total power is less than 350W. Many PSU's that look to have enough power may not have enough current for high level of overclocking. This number may vary a little bit (not a whole lot) due to component different. Detailed calculation in another thread. Look up the current spec for the good PSU's with enough current rating and small line regulation (e.g. 3% is better than 5%).

The Vcore regulator (those 4 or 6 big MOSFET's near the CPU socket are part of the 2-phase or 3-phase regulator) on mb becomes important when overclocking to high clock frequency (2.5+ GHz) which requires large active current. Mb that derives Vcore from 12V PSU line has better current sourcing capability, I think. Sometimes max oc may be limited by the Vcore regulator stability and cooling, this is something that may be overlooked. I found that good case air flow over these MOSFET is a must for Vcore stability. The MOSFET can benefit from the cool air coming from the CPU fan overhang of a high CFM fan such as TT SFII over a good HS such as SLK-800U/947U.


2. FSB clock frequency: Mb such as those w/ nforce2 chipset, PCI freq is locked, they can go at least to 180 MHz. If you are getting new mb currently, mb with nforce2 chipset is a good choice, new rev 2 mb with C1 chipset stepping supports least 200 MHz officially.

FSB and memory bus frequeny are locked at a ratio. If the ratio = 1 or 100% it is called SYNC mode, and is recommended for mb such as nforce2. If the ratio != 1, it is called ASYNC mode where FSB and memory bus are running at different frequency but in locked steps.

How high can FSB go depends on
- memory rated speed
- chipset quality (stepping) and chipset Vdd
- chipset (both NB and SB) cooling

To get higher FSB, make sure the memory module is rated to handle the target FSB at a given ras/cas timing and voltage. E.g. PC3200 memory will support to 200 MHz FSB (or DDR400), PC3500 for 217 MHz (DDR 434), ...

For these RAS/CAS memory timing such as 6-2-2-2 1T or 6-3-3-2 1T. They refer to:
1. Active (to) Precharge Delay (aka Tras, tRAS) - usually 5, 6, 7, ... (Tras >= Trcd + CAS)
2. RAS to CAS Delay (aka Trcd, tRCD) - 2 is good, 3 is OK
3. RAS Precharge Delay (aka Trp, tRP, Precharge to Active) - 2 is good, 3 is OK
4. CAS Latency (aka CAS) - use 2 whenever possible
5. Cmd Rate (some bios does not have this, set automatically) - 1T is better than 2T

Memory timing should be set to 6-3-3-2 to begin with to get to as high a FSB as possible, then tighten it to x-3-2-2 or x-2-2-2 (where x = 5, 6), if possible but not necessary, after reaching max FSB. The difference between 6-3-3-2 and x-2-2-2 is minor (~2%) on overall system performance.

Regarding to SYNC (memory bus speed = FSB) vs ASYNC (otherwise), IMO, running slower memory in ASYNC at 50-66-75% is much more price/performace effective than 100% SYNC. 50% memory cost will get to 10-15% of the max bandwidth. This approach has been used in P4 dual channel since its FSB is QDR (quad pump data) and there is no fast memory to match at that speed, and dual channel is the only way to fill up the system bandwidth. But for AMD system which is DDR, in order to get absolute performance, one will have to use the fastest memory to match the FSB and run them in SYNC to get the last 10% of memory bandwidth.

ASYNC is a good feature for testing of the system. For example, if you have slower memory (for the time being), and you want to test whether the system can run at higher FSB, you can set them to fsb:memory = 4:3, so memory running at 166 MHz and FSB at 221 MHz. Or vice versa, if you want to test you fast memory (e.g. PC 3500 rated to run at 217 MHz), and you motherboard can only run at 200 MHz, you can set the fsb:memory = 3:4, so FSB = 160 MHz and memory at 213 MHz. Then you can adjust FSB at 1 MHz at a time above 160 MHz to see whether the memory can indeed run at spec of 217 MHz or beyond. You can use memtest86 to test the memory in ASYNC this way.

Dual channel or single channel mode in nforce2 mb is not that crucial for overall performance. The difference is only few % (say 2-3%) at most. Also single channel may let FSB to go a bit higher due to a smaller chance of potential dual dimm mismatch and memory controller stress at high FSB, I think. On the other hand, dual channel memory controller provides some performance advantage due to its intrinsic speculative caching capability. At this point, the little higher FSB from single channel offset the performance advantage of dual channel, and the two is about a tie, I think, for AMD mb. For some nforce2 mb that have integrated video which can benefit from twice the nforce2 memory bandwidth, since the bus between the video and the memory controller has 2x64 bit bus.

The max bandwdith between memory controller and CPU would be 2 x 8 x FSB = 16 FSB MB/s. x2 is because of DDR (data are transferred at both rising and falling edge of the FSB clock, x8 because of 8-byte bus or 64-bit bus). The effective bandwidth, taking into memory controller (~95% efficiency), would be around 15.2 FSB. E.g. FSB = 200 MHz, effective bandwidth ~ 3040 MB/s.

Dual channel makes a big difference for P4 dual channel mb though, due to quad pump data of P4 (or QDR). The max bandwidth for P4 dual channel is 4 x 8 x FSB = 32 FSB MB/s. The effective bandwidth, taking into memory controller overhead (~ 75% efficiency), would be around 24 FSB MB/s. E.g. FSB = 200 MHz, effective bandwidth ~ 4800 MB/s, which is around 60% more than that of a nforce2 mb running same FSB 200 MHz. E.g. running fsb:memory=5:4, with FSB=250, memory=200, effective bandwidth ~ 24 x 225 = 5400 MB/s.

Chipset Vdd and Cooling on FSB stability

If wanting to go higher, chipset Vdd and cooling may have to be looked into. But this is secondary compared to getting the correct memory module that is rated to run at the desired FSB frequency and timing.

Some new mb, such as NF7-S rev 2, has wide range voltage for chipset Vdd, Vcore, Vdimm, so no hardware Vmod is needed. Otherwise goto the individual mb section, there are lots of info about various Vmod.

Putting a passive heat sink on the SB with good air flow over it should improve stability at high FSB oc. Adding a small fan on the heat sink of the NB (if there is no) and SB would be even better if wanting to go even higher FSB (220+ MHz). Good case air flow over these NB and SB chips are recommended and should be enough for high FSB (at least up to 230 MHz) stability. For air cooling, if CPU has high CFM fan attached (e.g. TTSFII over SLK-800U/947U), cool air from the fan overhang may help to cool the NB and also the MOSFET of the voltage regulator.

Nforce2 rev 2.0 motherboards with A1 stepping chipset (aka 400 Ultra) do not require heavy NB cooling and high chipset Vdd as the old rev 1.0 did. I found that for up to 230 FSB, stock HSF and 1.7 V Vdd from bios setting (w/o Vmod) should be enough.

RAID: One cons for recent nforce2 motherboard is that there is no direct RAID support for the traditional IDE parallel hard drives. RAID-0 delivers exceptional hard disk performance on older KT266/333 motherboards. The new serial hard drives are still not common and expensive. One can get two parelllel-to-serial converters to connect existing IDE (parallel) HD to the onboard SATA interface which supports only RAID-0 or RAID-1, but not for RAID-0+1.

Memory testing

Test the memory module using memtest95 at the rated frequency and voltage to make sure it runs at spec. memtest95 is running almost on bare hardware, outside of Windows, so the testing can be done without some uncertainty and stability issues related to chipset drivers, video drivers and intensive FSB traffic, ...

Another nice feature of new mb is that FSB and memory bus speed can be run ASYNC with each other, for memory and FSB testing to avoid having uncertainty for both of them. Set one of them slower and oc the other as much as possible to find out their limit separately. E.g. set fsb:memory ratio at 3/4, would allow you to test memory at 200 MHz while running FSB for sure at 150 MHz, etc, etc, ...

Summary

Even you have a motherboard running at slow FSB (133/166 MHz), you can still overclock the CPU to highest possible frequency, using high mulitpliers (assuming multiplier limit is resolved). That is, max CPU overclock frequency is independent of FSB. Max FSB overclocking does not depend on the CPU per se, since the multiplier can resolve that, assuming you have an unlocked CPU, such as Tbred. The best of both world is that you have a motherboard and memory that can achieve high FSB, and a CPU that can be overclocked to very high freqency.

Assuming you have a recent nforce2 rev. 2 mothboard, an highly overclockable CPU such as Tbred B 1700+ DLT3C, Barton 2500+, ... The nforce2 rev. 2 motherboard (such as NF7-S) is a good choice, since it has a wide range of voltage for Vcore, Vdmm (for memory), Vdd (for chipset FSB overclocking), Vagp than others, so Vmod is not needed in general. Further it uses 12V and 3-phase regulator both of which IMO are better for generate stable Vcore. If you have good heat sink (e.g. SK7 or SLK-800/900), adjustable high CFM fan, good PSU w/ enough active current sourcing rails and lines with small regulation (3% prefer), good CAS2 DDR PC3500 memory, then they are sufficient to get the CPU and FSB to above average overclocking, within 5% of the full potential of your system.

The last 3-5% would be more costly and may require more experience and technique and possibly better components. But probably at that point, one would have already acquired enough know how to get to the max oc and beyond.

=============================================

Adjusting FSB

The ideal situation is that both the FSB and CPU clock frequency are highest possible. FSB should be highest possible so system bandwidth (between CPU and north bridge for XP and P4) for memory and video are highest. Then CPU clock frequency would following FSB x multiplier by adjusting the multiplier. There is also some fine tuning at the end between the exact FSB (down to 1 MHz) and multiplier (down to +- 0.5), so as to tradeoff the final % between the FSB and CPU, since the exact CPU frequency may not be at the exact quantized level given by FSB.

At the beginning, keep the FSB close to its max (then the mulitplier has to be set to low accordingly, since multiplier = CPU_frequency/FSB). This number depends on your motherboard and memory. For recent rev 2 nforce2 mb, and if you have PC3200 CAS2 memory or better, 200 MHz is a good starting point. Then increase mulitplier to locate roughly the CPU max while increase Vcore 25 mV at a time. Run prime95 to check for stability.

Near highest overclocking, mulitplier is kept constant (or increase/decrease multiplier by 0.5 if needed, to allow room for FSB to change). FSB is increased/decreased 1 MHz or small steps (trial and error) until either the FSB or the CPU (or both but not likely) are maxed out (system unstable) for a given Vcore setting. The 3DMark01, 3DMark03 benchmark programs can be used to test the overall system stability, stressing FSB, memory, video in particular.

Sandra is a commonly used program for benchmarking CPU integer and floating point operations raw computing power in DMIPS and FLOPS, memory bandwidth, cache performance, multimedia performance, ... as well as some informations about system hardwares and devices.

For overclocking, don't use default setting. Set them to manual, so you can adust the FSB, mulitplier, memory timing (6-3-3-2 is a good starting point), ...

How to adjust Vcore

After finding the max FSB (or within 5-10 MHz), for CPU overclocking, CPU clock is increased (bigger step at beginning and few MHz step at the end). And when the system becomes unstable, you can increase Vcore by 25 mV stepwise to see whether it can make the CPU to run faster so that the system can become stable again.

Repeat the above until high Vcore is reached to a point that CPU clock frequency cannot go higher practically (say less than 5-10 MHz while temperature rises by 5C per step of Vcore (25 mV)), reaching zone of diminishing return on frequency. From experience, 1.9-1.95 V should be OK for typical Tbred B overclocking (at least for testing), system stability (e.g. running Prime95 for CPU overclocking) is key, while keeping die temperature under control (take 50-55 C loaded as reference rather than the perceived absolute limit). Good HSF is needed to keep temperature low.

There are different level of stability. Minimum is that it can boot in the operating system and run Sandra CPU. Prime95 is considered to be a reasonable way for CPU stability and stress test.

Overclocking is a process, not by sudden boosting the Vcore and hope it will go to 2.5 GHz. Then it would not blow up the CPU. Don't sudden jump Vcore by multiple steps since you will lose sight of the CPU stability and die temperature increases.

During this process, watch for
- system stability,
- rate of increase of clock frequency per step of Vcore increase
(I consider 5-10 MHz per 25 mV Vcore increase as near the overclocking limit)
- die temperature, keep an eye on the rate of temperature increase per step of Vcore increase
(I consider 5 C increase for loaded CPU per 25 mV step of Vcore increase as near the overclocking limit), take 50-55-65 C as relative rather than perceived absolute limit
- By combining clock frequency and temperature, the rule of thumb is:
10 MHz / C, at full load per 25 mV Vcore increase is about the break-even point for overclocking (point of diminishing return). Below 10 MHz / C is "easy" and economical for overclocking, above 10 MHz / C becomes harder and harder to overclock. And 2-3 MHz / C at full load is the limit.


Example for Vcore increase

Here use a Tbred B 1700+ DLT3C as an exmaple. If you have a CPU that can do 2.2 GHz @ 1.5V, it is a good CPU for oc.
(If you chip can only do 2.1 GHz @ 1.5V, then substract the CPU clock by 100 MHz and 0.5 in the multiplier in the following example, etc, etc.)

Keep FSB = 200 MHz
Raise the multiplier to 11.5, i.e. freq = 2300 MHz
Increase Vcore to 1.6 V+- should get it stable. Run prime95 to test stability.

Then repeat again with FSB = 200 MHz, multiplier 12. I estimate Vcore around 1.7 V +- to get it stable to 200 x 12 = 2400 MHz. Run prime95 to test stability.
etc, etc.

At the early stage, one would expect to get 100+ MHz for each 0.1 V or 100 mV Vcore increase. When close to the overclocking limit, that number will drop to around 50 MHz / 100 mV, and eventually down to 5-10 mV / 25 mV eventually, and the end is almost there. Also at the last % of overclocking, the temperature would increase much faster per step of Vcore increase, I consider 5 C/ 25 mV is about reaching the limit. It will be very costly in terms of voltage, power and temperature to get the CPU to run a % higher.

This is just an estimated scenario, try it out first. If you want to go further, you may need to try a better HSF such as SK7 or SLK-800U/900U with a variable speed high CFM fan such as TT SFII or even a Vantec Tornado (with VR mod).

When you can reach 2.4/2.5 GHz, depends on the HSF and PSU, your are probably 100 MHz away from one of the best oc of a 1700+ on air.

I've set my limits at the low 50s and 1.65v

I'll overclock the best I can under those limits.

i seem to be running prime fine on 1.950volts at 200*11.5 @ 48C (full load) {barton} meaning its rated like 2.5 or so on a tbred

this is the first day i got this cpu.... its gonna take some burning in before i can get the cpu to lower volts, and lower temps

hitechjb1 said:

We should arrange a lan oc party in the south pole, or move to the south pole for oc during the summer time in the northern hemisphere.

Click to expand...

lol, overclocking migration now thats new HAHAHA

I have been having pretty good results with mine aswell. THis is with a tornado and slk-800U. I will try for more once i burn it in

Nice result on air, Congrats.

What stepping, week of the CPU ?

What PSU do you use?

hitechjb1

i just noticed that your mem scores in pcmark were a couple hundred points below mine, but you are at 209-211fsb and i'm at 195

so i'm wondering what timings you are using

its a 310 xpmw and its on a crappy antec 400w, i should be ordering a new True Control 550 this week though

specific said:

hitechjb1

i just noticed that your mem scores in pcmark were a couple hundred points below mine, but you are at 209-211fsb and i'm at 195

so i'm wondering what timings you are using

Click to expand...

I am using one generic 512MBx1 PC3500C2 module, setting at 5-3-2-2 2.6V. That is the tighest timing it can run. Sandra memory bandwdith number was 3144 MB/s at 210 MHz.

Do you have better memory timing? What sandra memory score?

WTF 33C? hmmm are you sure you got a SLK 800 on there?
does it help that much? i got a barton now at full load its around 48 with a volcano 9.... (200*11.5)~2.3002

EspElement said:

WTF 33C? hmmm are you sure you got a SLK 800 on there?
does it help that much? i got a barton now at full load its around 48 with a volcano 9.... (200*11.5)~2.3002

Click to expand...

Probably that temp is idle, I guess..


Comparing SLK-800 with Volcano 9

Yes, the SLK-800 would help a lot, here is the estimate to show.

When fan at around 5500 rpm,
The thermal resistance for SLK-800 is 0.23 C/W
The thermal resistance for SK7 is 0.25 C/W
The thermal resistance for Volcano 9 is 0.36 C/W
(thermal resistance number from heat sink page of this forum).

If a Tbred B 1700+ DLT3C is oc'ed to 2.5 GHz, 1.9V, active thermal power = 136 W (2.7 time default !!)

the die temperature difference between the two HS = 136 x (0.36 - 0.23) = 17.7 C !!!!

That is why no doubt for high oc for high power CPU, SK7 (economical) or SLK-800 or SLK-900 is needed w/ a high speed fan also (for pics).

That's an incredibly low temp, even idle, for that high of a clock speed. Usually full load temps go about 5° C or so higher than idle. Probably just a case of inaccurate thermistor readings.